brainy/src/db/db.ts

1041 lines
44 KiB
TypeScript
Raw Normal View History

feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* @module db/db
* @description `Db` Brainy 8.0's immutable, Datomic-style database value.
*
* A `Db` is a **readonly view pinned at one generation** of a Brainy store.
* It is produced by `brain.now()` (O(1) pin of the current generation),
* `brain.transact()` (pinned at the freshly committed generation),
* `brain.asOf()` (a past generation, a timestamp, or a persisted snapshot
* path), `Brainy.load(path)` (a snapshot opened read-only), and `db.with()`
* (a speculative in-memory overlay).
*
* **Read semantics.** While no transaction has committed past the pinned
* generation, every read delegates straight to the live brain the pinned
* view IS the current state, and the existing fast paths serve it untouched.
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* Once later transactions commit, the `Db` serves the FULL query surface at
* the pinned generation through two complementary paths:
*
* - `get()`, metadata-level `find()`, and filter-based `related()` resolve
* through the generational record layer (changed ids from immutable
* before-images; unchanged ids still ride the live fast path) no
* materialization cost.
* - Index-accelerated queries (semantic/vector search, graph traversal,
* cursors, aggregation) are served by **at-generation index
* materialization**: on first use, the host rebuilds ephemeral in-memory
* indexes (the same vector/metadata/graph index classes the live brain
* uses) over the exact at-generation record set, caches them on this `Db`,
* and frees them on {@link Db.release}. This costs O(n at G) time and
* memory ONCE per `Db` the open-core price of historical index queries;
* a native {@link ../plugin.js VersionedIndexProvider} serves the same
* reads from its retained segments without any rebuild.
*
* Speculative `with()` overlays keep one honest boundary: overlay entities
* carry no embeddings, so index-accelerated queries on overlays throw
* {@link SpeculativeOverlayError} instead of returning silently-incomplete
* results.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
* **History granularity.** Generation records are written per `transact()`
* batch. Single-operation writes (`add`/`update`/`relate`/ outside
* `transact()`) advance the generation counter but do not produce historical
* records, so they remain visible through earlier pins the documented 8.0
* granularity (see `docs/ADR-001-generational-mvcc.md`, "History
* granularity").
*
* **Lifecycle.** Each `Db` holds one refcounted pin on its generation (and
* on every registered {@link ../plugin.js VersionedIndexProvider}). Call
* {@link Db.release} when done a `FinalizationRegistry` backstop releases
* leaked pins on garbage collection, but explicit release is what makes
* `compactHistory()` deterministic.
*/
import type {
Entity,
FindParams,
GetOptions,
RelatedParams,
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
Relation,
Result
} from '../types/brainy.types.js'
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
import type { StorageAdapter } from '../coreTypes.js'
refactor(8.0): rename BackupData → PortableGraph (the type is interchange, not a backup) The export()/import() document type was named BackupData, but it is not a backup: it is the portable, versioned, partial-or-whole interchange representation of a graph (entities + relations + optional vectors/blobs) — the unit Brainy exports for transport between instances, versions, and products, and the payload other artifacts embed. The actual backup is persist()/load() (the native whole-brain, generation-preserving snapshot), so "Backup*" actively collided with that concept. Rename every developer-visible symbol, file, doc, JSDoc and comment: - BackupData→PortableGraph, BackupEntity→PortableGraphEntity, BackupRelation→PortableGraphRelation, BackupReader/Writer→PortableGraphReader/Writer, BackupValidation→PortableGraphValidation, isBackupData→isPortableGraph, validateBackup→validatePortableGraph, BACKUP_FORMAT[_VERSION]→PORTABLE_GRAPH_FORMAT[_VERSION]. - src/db/backup.ts → src/db/portableGraph.ts; test → db-portable-graph.test.ts. - Guide, api/README, RELEASES updated. The on-the-wire `format` tag is renamed 'brainy-backup' → 'brainy-portable-graph': the format was introduced in 7.32.0 and has not been adopted by any consumer, so there are no stored documents to stay compatible with — a clean rename beats carrying a legacy tag. No deprecated aliases (nothing to alias). 7.x shipped the same rename as 7.32.2. 1471 unit green; build clean; db-portable-graph.test.ts 17/17.
2026-06-19 12:09:19 -07:00
import { exportGraph } from './portableGraph.js'
import type { ExportSelector, ExportOptions, PortableGraph } from './portableGraph.js'
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
import {
splitNounMetadataRecord,
splitVerbMetadataRecord
} from '../types/reservedFields.js'
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
import { v4 as uuidv4 } from '../universal/uuid.js'
import { coerceNewEntityId, resolveEntityId, ORIGINAL_ID_KEY } from '../utils/idNormalization.js'
import { EntityNotFoundError } from '../errors/notFound.js'
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
import { SpeculativeOverlayError } from './errors.js'
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
import type { GenerationStore } from './generationStore.js'
import type { ChangedIds, TransactReceipt, TxOperation } from './types.js'
import { entityMatchesFind, resolveEntityField, UnsupportedWhereOperatorError } from './whereMatcher.js'
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/**
* @description The query surface a historical materialization exposes back
* to its `Db`: the full live read pipeline of an ephemeral reader brain
* whose storage holds the exact at-generation record set. Produced by
* {@link DbHost.materializeAt}; closed (indexes freed) via `close()` when
* the owning `Db` is released.
*/
export interface HistoricalQueryHandle<T = any> {
/** Full `find()` surface (vector/semantic/graph/cursor/aggregate) at the pinned generation. */
find(query: string | FindParams<T>): Promise<Result<T>[]>
/** Full `related()` surface (including cursor pagination) at the pinned generation. */
related(paramsOrId?: string | RelatedParams): Promise<Relation<T>[]>
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/** Free the materialized indexes (closes the ephemeral reader). Idempotent. */
close(): Promise<void>
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* @description The speculative overlay carried by a `db.with()` value:
* per-id replacement entities/relations (`null` = tombstone). Reads on the
* overlay `Db` consult these maps first, then fall through to the pinned
* base view. Overlays are pure in-memory values they never touch disk or
* index providers.
*/
export interface SpeculativeOverlay<T = any> {
/** Entity overlays: replacement entity, or `null` for a speculative delete. */
nouns: Map<string, Entity<T> | null>
/** Relation overlays: replacement relation, or `null` for a speculative delete. */
verbs: Map<string, Relation<T> | null>
}
/**
* @description The host surface a `Db` needs from its `Brainy` instance
* constructed once per brain (see `brainy.ts`). Internal: consumers never
* build one; `Db` values are only created by `Brainy` and `db.with()`.
*/
export interface DbHost<T = any> {
/** The generational record layer of the host brain. */
readonly store: GenerationStore
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
/** The host brain's storage adapter (portable-export VFS blob bytes). */
readonly storage: StorageAdapter
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/** Live `brain.get()` (current-generation fast path). */
get(id: string, options?: GetOptions): Promise<Entity<T> | null>
/** Live `brain.find()` (current-generation fast path). */
find(query: string | FindParams<T>): Promise<Result<T>[]>
/** Live `brain.related()` (current-generation fast path). */
related(paramsOrId?: string | RelatedParams): Promise<Relation<T>[]>
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
/**
* Resolve a `generation | Date` to a concrete generation (+ timestamp) via the
* brain's shared `asOf` resolver — so `since(gen|Date)` shares `asOf`'s
* reachability and `Date` semantics. `exclusive` pins the generation before the
* resolved one.
*/
resolveGeneration(
target: number | Date,
options?: { exclusive?: boolean }
): Promise<{ generation: number; timestamp: number }>
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/** Materialize an entity from a generation record's raw stored objects. */
entityFromRecord(
id: string,
record: { metadata: any; vector: any | null },
includeVectors: boolean
): Promise<Entity<T>>
/** Materialize a relation from a generation record's raw stored objects. */
relationFromRecord(id: string, record: { metadata: any; vector: any | null }): Relation<T> | null
/** Snapshot the store at `generation` into `targetPath` (throws if the pin is stale). */
persistPinned(targetPath: string, generation: number): Promise<void>
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/**
* Build the at-`generation` index materialization: an ephemeral in-memory
* reader over the exact record set at that generation, serving the full
* query surface. O(n at G) called at most once per `Db` (cached by the
* caller) and freed via the returned handle's `close()`.
*/
materializeAt(generation: number): Promise<HistoricalQueryHandle<T>>
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/** Add one refcounted pin (store + versioned providers) on `generation`. */
pinGeneration(generation: number): void
/** Release one refcounted pin (store + versioned providers) on `generation`. */
releaseGeneration(generation: number): void
/** Register a `Db` with the GC-backstop finalization registry. */
registerDbForFinalization(db: Db<T>, generation: number, closeOnRelease?: () => Promise<void>): void
/** Unregister a `Db` after an explicit release (no double-release on GC). */
unregisterDbFromFinalization(db: Db<T>): void
}
/**
* @description Internal constructor arguments for {@link Db}. The pin on
* `generation` is taken by the creator (`Brainy` or `db.with()`) **before**
* construction; the constructor only registers the GC backstop.
*/
export interface DbInit<T = any> {
/** The host surface (one per brain). */
host: DbHost<T>
/** The pinned generation. */
generation: number
/** The view's timestamp (ms since epoch) — see {@link Db.timestamp}. */
timestamp: number
/** Per-operation receipt, present only on `transact()`-produced values. */
receipt?: TransactReceipt
/** Speculative overlay, present only on `with()`-produced values. */
overlay?: SpeculativeOverlay<T>
/** Extra teardown on release (e.g. closing an owned snapshot brain). */
closeOnRelease?: () => Promise<void>
}
/**
* @description An immutable, generation-pinned view of a Brainy store. See
* the module documentation for the full read-semantics contract.
*
* @example
* const db = brain.now() // O(1) pin
* await brain.transact([{ op: 'update', id, metadata: { v: 2 } }])
* await db.get(id) // still sees v: 1 — pinned
* await brain.get(id) // sees v: 2 — live
* await db.release() // unpin (enables compaction)
*/
export class Db<T = any> {
private readonly host: DbHost<T>
private readonly gen: number
private readonly ts: number
private readonly overlay?: SpeculativeOverlay<T>
private readonly closeOnRelease?: () => Promise<void>
private isReleased = false
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/**
* Cached at-generation index materialization (built lazily by the first
* index-accelerated query at a historical pin; freed on release()).
*/
private materializedHandle?: Promise<HistoricalQueryHandle<T>>
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* Per-operation receipt (committed generation, timestamp, resolved ids in
* input order). Present only on values returned by `brain.transact()`.
*/
public readonly receipt?: TransactReceipt
/**
* @param init - Internal construction arguments. `Db` values are created
* by `Brainy` (`now`/`transact`/`asOf`/`load`) and `db.with()` never
* directly by consumers.
*/
constructor(init: DbInit<T>) {
this.host = init.host
this.gen = init.generation
this.ts = init.timestamp
this.receipt = init.receipt
this.overlay = init.overlay
this.closeOnRelease = init.closeOnRelease
this.host.registerDbForFinalization(this, this.gen, this.closeOnRelease)
}
/** The generation this view is pinned at. */
get generation(): number {
return this.gen
}
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
/**
* @internal Same-instance guard for cross-`Db` operations like
* `Brainy.diff()`: whether this view is backed by `store`. Lets `Brainy`
* reject a `Db` endpoint from a different brain without exposing the private
* host.
*/
belongsToStore(store: GenerationStore): boolean {
return this.host.store === store
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* The view's wall-clock timestamp (ms since epoch): pin time for `now()`,
* commit time for `transact()`, and the resolved commit time for `asOf()`.
*/
get timestamp(): number {
return this.ts
}
/** Whether {@link Db.release} has been called (released views throw on use). */
get released(): boolean {
return this.isReleased
}
/** Whether this view carries a speculative `with()` overlay. */
get speculative(): boolean {
return this.overlay !== undefined
}
// ==========================================================================
// Reads
// ==========================================================================
/**
* @description Get an entity by id **as of this view's generation**.
* Overlay entries win (speculative views); ids untouched since the pin ride
* the live fast path; ids changed by later transactions resolve from
* immutable generation records. Always correct at the pinned generation.
*
* @param id - The entity id.
* @param options - Same options as `brain.get()` (`includeVectors`).
* @returns The entity as of this generation, or `null`.
*/
async get(id: string, options?: GetOptions): Promise<Entity<T> | null> {
this.assertUsable('get')
// Id normalization (8.0): resolve a natural key to the canonical UUID the
// write path stored, so this view reads by natural key consistently. A real
// UUID passes through. Overlay keys are canonical (see with()).
id = resolveEntityId(id)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (this.overlay && this.overlay.nouns.has(id)) {
return this.overlay.nouns.get(id) ?? null
}
if (!this.isHistorical()) {
return this.host.get(id, options)
}
const resolved = await this.host.store.resolveAt('noun', id, this.gen)
if (resolved.source === 'current') {
return this.host.get(id, options)
}
if (resolved.source === 'absent' || resolved.metadata === null) {
// Live semantics: an entity without metadata is not a live entity.
return null
}
return this.host.entityFromRecord(
id,
{ metadata: resolved.metadata, vector: resolved.vector },
options?.includeVectors ?? false
)
}
/**
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* @description The full `find()` surface **as of this view's generation**.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* Routing:
* - **Current generation, no overlay** delegates to `brain.find()`
* untouched (the existing fast paths).
* - **Historical generation** metadata dimensions (`type`, `subtype`,
* `where`, `service`, `excludeVFS`, `orderBy`/`order`,
* `limit`/`offset`) are computed from the generational record layer
* directly (no materialization cost): the live index serves unchanged
* ids, and per-entity evaluation of the same filters covers
* record-resolved ids set-correct at the pinned generation.
* Index-accelerated dimensions (`query`, `vector`, `near`, `connected`,
* `cursor`, `aggregate`, `includeRelations`, non-metadata search modes)
* are served by the at-generation index materialization built lazily
* on first use (O(n at G) time and memory, once per `Db`; see the module
* doc), then cached until {@link Db.release}.
* - **Speculative overlay** metadata dimensions only; index-accelerated
* dimensions throw {@link SpeculativeOverlayError} (overlay entities
* carry no embeddings see the error's documentation).
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
* Result ordering is deterministic when `orderBy` is supplied; without it,
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* record-path results list live-index matches before record-resolved
* matches, while materialized results follow the live engine's ordering.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* @param query - A `FindParams` object, or a string (semantic search).
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* @returns Matching results as of this generation (score `1.0` for
* record-resolved metadata matches, mirroring live metadata-only finds).
*/
async find(query: string | FindParams<T>): Promise<Result<T>[]> {
this.assertUsable('find')
const params: FindParams<T> = typeof query === 'string' ? { query } : query
const historical = this.isHistorical()
if (!historical && !this.overlay) {
return this.host.find(query)
}
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
if (this.overlay) {
this.assertOverlayCompatibleFind(params)
} else if (findRequiresIndexes(params)) {
const materialized = await this.materialize()
return materialized.find(params)
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
const limit = params.limit ?? 10
const offset = params.offset ?? 0
// Ids whose live-index answer is NOT valid at this generation.
const changedNouns = historical
? (await this.host.store.changedBetween(this.gen, this.host.store.committedGeneration())).nouns
: []
const overlayNouns = this.overlay?.nouns ?? new Map<string, Entity<T> | null>()
const excluded = new Set<string>([...changedNouns, ...overlayNouns.keys()])
try {
// Over-fetch from the live index so dropping superseded ids cannot
// starve the requested page, then merge and re-window.
const base = await this.host.find({
...params,
limit: limit + offset + excluded.size,
offset: 0
})
const merged = base.filter((result) => !excluded.has(result.id))
for (const id of changedNouns) {
if (overlayNouns.has(id)) continue // The overlay supersedes history.
const resolved = await this.host.store.resolveAt('noun', id, this.gen)
let entity: Entity<T> | null = null
if (resolved.source === 'record' && resolved.metadata !== null) {
entity = await this.host.entityFromRecord(
id,
{ metadata: resolved.metadata, vector: resolved.vector },
false
)
} else if (resolved.source === 'current') {
// Defensive: changed ids always resolve to a record or absent, but
// a 'current' answer is still served correctly from the live path.
entity = await this.host.get(id)
}
if (entity && entityMatchesFind(entity as Entity, params as FindParams)) {
merged.push(resultFromEntity(entity))
}
}
for (const [, entity] of overlayNouns) {
if (entity === null) continue // Speculative delete.
if (entityMatchesFind(entity as Entity, params as FindParams)) {
merged.push(resultFromEntity(entity))
}
}
if (params.orderBy) {
sortResultsBy(merged, params.orderBy, params.order ?? 'asc')
}
return merged.slice(offset, offset + limit)
} catch (err) {
if (err instanceof UnsupportedWhereOperatorError) {
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
// The record-path's per-entity matcher doesn't implement this
// where-operator. On an overlay that is a hard boundary; at a
// historical generation the materialization serves it via the full
// live query engine.
if (this.overlay) {
throw new SpeculativeOverlayError(
`metadata find with where-operator '${err.operator}'`,
this.gen
)
}
const materialized = await this.materialize()
return materialized.find(params)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
throw err
}
}
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
/**
* @description Serialize part or all of this database value into a portable
refactor(8.0): rename BackupData → PortableGraph (the type is interchange, not a backup) The export()/import() document type was named BackupData, but it is not a backup: it is the portable, versioned, partial-or-whole interchange representation of a graph (entities + relations + optional vectors/blobs) — the unit Brainy exports for transport between instances, versions, and products, and the payload other artifacts embed. The actual backup is persist()/load() (the native whole-brain, generation-preserving snapshot), so "Backup*" actively collided with that concept. Rename every developer-visible symbol, file, doc, JSDoc and comment: - BackupData→PortableGraph, BackupEntity→PortableGraphEntity, BackupRelation→PortableGraphRelation, BackupReader/Writer→PortableGraphReader/Writer, BackupValidation→PortableGraphValidation, isBackupData→isPortableGraph, validateBackup→validatePortableGraph, BACKUP_FORMAT[_VERSION]→PORTABLE_GRAPH_FORMAT[_VERSION]. - src/db/backup.ts → src/db/portableGraph.ts; test → db-portable-graph.test.ts. - Guide, api/README, RELEASES updated. The on-the-wire `format` tag is renamed 'brainy-backup' → 'brainy-portable-graph': the format was introduced in 7.32.0 and has not been adopted by any consumer, so there are no stored documents to stay compatible with — a clean rename beats carrying a legacy tag. No deprecated aliases (nothing to alias). 7.x shipped the same rename as 7.32.2. 1471 unit green; build clean; db-portable-graph.test.ts 17/17.
2026-06-19 12:09:19 -07:00
* `PortableGraph` document, read **as of this view's generation**. Because `export`
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
* lives on the immutable `Db`, it composes with every way of obtaining one:
* `brain.now().export(sel)` (current), `(await brain.asOf(g)).export(sel)`
* (time-travel export), and `brain.now().with(ops).export(sel)` (what-if export).
*
* The document is portable JSON, versioned (`formatVersion`), and current-state
* (no generation history) distinct from `persist()` (native whole-brain snapshot
* that preserves history). Restore with `brain.import(backup)`.
*
* @param selector - WHAT to export (omit for the whole brain). See {@link ExportSelector}.
* @param options - HOW to export (vectors / VFS bytes / edge policy). See {@link ExportOptions}.
refactor(8.0): rename BackupData → PortableGraph (the type is interchange, not a backup) The export()/import() document type was named BackupData, but it is not a backup: it is the portable, versioned, partial-or-whole interchange representation of a graph (entities + relations + optional vectors/blobs) — the unit Brainy exports for transport between instances, versions, and products, and the payload other artifacts embed. The actual backup is persist()/load() (the native whole-brain, generation-preserving snapshot), so "Backup*" actively collided with that concept. Rename every developer-visible symbol, file, doc, JSDoc and comment: - BackupData→PortableGraph, BackupEntity→PortableGraphEntity, BackupRelation→PortableGraphRelation, BackupReader/Writer→PortableGraphReader/Writer, BackupValidation→PortableGraphValidation, isBackupData→isPortableGraph, validateBackup→validatePortableGraph, BACKUP_FORMAT[_VERSION]→PORTABLE_GRAPH_FORMAT[_VERSION]. - src/db/backup.ts → src/db/portableGraph.ts; test → db-portable-graph.test.ts. - Guide, api/README, RELEASES updated. The on-the-wire `format` tag is renamed 'brainy-backup' → 'brainy-portable-graph': the format was introduced in 7.32.0 and has not been adopted by any consumer, so there are no stored documents to stay compatible with — a clean rename beats carrying a legacy tag. No deprecated aliases (nothing to alias). 7.x shipped the same rename as 7.32.2. 1471 unit green; build clean; db-portable-graph.test.ts 17/17.
2026-06-19 12:09:19 -07:00
* @returns A versioned, portable `PortableGraph` document.
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
* @example
* const backup = await brain.now().export({ collection: id }, { includeVectors: true })
*/
refactor(8.0): rename BackupData → PortableGraph (the type is interchange, not a backup) The export()/import() document type was named BackupData, but it is not a backup: it is the portable, versioned, partial-or-whole interchange representation of a graph (entities + relations + optional vectors/blobs) — the unit Brainy exports for transport between instances, versions, and products, and the payload other artifacts embed. The actual backup is persist()/load() (the native whole-brain, generation-preserving snapshot), so "Backup*" actively collided with that concept. Rename every developer-visible symbol, file, doc, JSDoc and comment: - BackupData→PortableGraph, BackupEntity→PortableGraphEntity, BackupRelation→PortableGraphRelation, BackupReader/Writer→PortableGraphReader/Writer, BackupValidation→PortableGraphValidation, isBackupData→isPortableGraph, validateBackup→validatePortableGraph, BACKUP_FORMAT[_VERSION]→PORTABLE_GRAPH_FORMAT[_VERSION]. - src/db/backup.ts → src/db/portableGraph.ts; test → db-portable-graph.test.ts. - Guide, api/README, RELEASES updated. The on-the-wire `format` tag is renamed 'brainy-backup' → 'brainy-portable-graph': the format was introduced in 7.32.0 and has not been adopted by any consumer, so there are no stored documents to stay compatible with — a clean rename beats carrying a legacy tag. No deprecated aliases (nothing to alias). 7.x shipped the same rename as 7.32.2. 1471 unit green; build clean; db-portable-graph.test.ts 17/17.
2026-06-19 12:09:19 -07:00
async export(selector: ExportSelector = {}, options: ExportOptions = {}): Promise<PortableGraph> {
feat(8.0): portable graph export()/import() (BackupData v1) — Db.export + polymorphic import Re-adds the portable graph round-trip on 8.0 (deleted with DataAPI), now on the immutable Db so it composes with the generational model. - src/db/backup.ts: BackupData v1 engine (identical wire format to the 7.32.0 line). exportGraph() reads through a Db at its pinned generation; importGraph() applies the whole backup as ONE atomic transact() (a single generation). - Db.export(selector?, options?) — read at this view's generation, so brain.asOf(g).export() is a time-travel export and brain.now().with(ops).export() a what-if export. - brain.export(selector?, options?) — sugar for now().export(). - brain.import() is polymorphic: a BackupData document -> graph round-trip; a file/buffer -> existing foreign-file ingestion (dispatched on the 'brainy-backup' tag; zero migration for ingestion callers). - Selectors (ids/collection/connected/vfsPath/predicate/whole) + edge policy (induced/incident/none) + includeVectors/includeContent/includeSystem; system entities excluded by visibility. onConflict merge/replace/skip, reembed auto/never, remapIds. DbHost gains storage (VFS blob bytes). - Types exported from the package root: BackupData/BackupEntity/BackupRelation/ ExportSelector/ExportOptions/ImportOptions/ImportResult + isBackupData. Tests: tests/unit/db/db-backup.test.ts (10, green) — round-trip, selectors, edges, vectors+re-embed, merge, asOf/with composition, validation. Full unit suite green. Follow-ups: docs (8.0 guide + RELEASES + api/README), completeness adds (since(prior).export() delta, exportStream/importStream, validate()), includeContent filesystem test, ids-at-asOf get() refinement, @soulcraft/formats Zod schema mirror.
2026-06-16 16:38:18 -07:00
this.assertUsable('export')
return exportGraph(this, this.host.storage, selector, options)
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* @description Relationships **as of this view's generation** the `Db`
* counterpart of `brain.related()` (same string-id shorthand and
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* filter surface). Relations whose edges changed after the pin are
* resolved from generation records; overlay relations (speculative
* `relate`/`unrelate`) and cascade tombstones (relations touching a
* speculatively deleted entity) are applied on top. Cursor pagination is
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* index-only: at a historical generation it is served by the
* at-generation index materialization (O(n at G) once per `Db`); on a
* speculative overlay it throws {@link SpeculativeOverlayError}.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
* @param paramsOrId - An entity id (shorthand for `{ from: id }`) or a
* `RelatedParams` filter object.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* @returns Relations as of this generation.
*/
async related(paramsOrId?: string | RelatedParams): Promise<Relation<T>[]> {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
this.assertUsable('related')
const rawParams: RelatedParams =
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
typeof paramsOrId === 'string' ? { from: paramsOrId } : (paramsOrId ?? {})
// Id normalization (8.0): resolve the anchor id(s) so this view traverses by
// natural key consistently with the live engine. Real UUIDs pass through.
const params: RelatedParams = {
...rawParams,
...(rawParams.from !== undefined && { from: resolveEntityId(rawParams.from) }),
...(rawParams.to !== undefined && { to: resolveEntityId(rawParams.to) })
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
const historical = this.isHistorical()
if (!historical && !this.overlay) {
return this.host.related(params)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
if (params.cursor !== undefined) {
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
if (this.overlay) {
throw new SpeculativeOverlayError('cursor pagination on related()', this.gen)
}
const materialized = await this.materialize()
return materialized.related(params)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
const limit = params.limit ?? 100
const offset = params.offset ?? 0
const changedVerbs = historical
? (await this.host.store.changedBetween(this.gen, this.host.store.committedGeneration())).verbs
: []
const overlayVerbs = this.overlay?.verbs ?? new Map<string, Relation<T> | null>()
const excluded = new Set<string>([...changedVerbs, ...overlayVerbs.keys()])
const base = await this.host.related({
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
...params,
limit: limit + offset + excluded.size,
offset: 0
})
let merged = base.filter((relation) => !excluded.has(relation.id))
for (const id of changedVerbs) {
if (overlayVerbs.has(id)) continue
const resolved = await this.host.store.resolveAt('verb', id, this.gen)
if (resolved.source !== 'record') continue
const relation = this.host.relationFromRecord(id, {
metadata: resolved.metadata,
vector: resolved.vector
})
if (relation && relationMatchesParams(relation, params)) {
merged.push(relation)
}
}
for (const [, relation] of overlayVerbs) {
if (relation === null) continue // Speculative unrelate.
if (relationMatchesParams(relation, params)) {
merged.push(relation)
}
}
// Cascade semantics for speculative deletes: a relation whose endpoint
// is tombstoned in the overlay is gone in this view, exactly as
// brain.remove() cascades on the durable path.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (this.overlay) {
const tombstoned = new Set<string>()
for (const [id, entity] of this.overlay.nouns) {
if (entity === null) tombstoned.add(id)
}
if (tombstoned.size > 0) {
merged = merged.filter(
(relation) => !tombstoned.has(relation.from) && !tombstoned.has(relation.to)
)
}
}
return merged.slice(offset, offset + limit)
}
// ==========================================================================
// Derived views
// ==========================================================================
/**
* @description Speculative write: returns a NEW `Db` whose reads see the
* given transaction data applied **in memory, on top of this view**
* Datomic's `with`. Nothing touches disk, the generation counter, or index
* providers; the underlying brain and this view are untouched. Use it to
* answer "what would the store look like if…" questions, then call
* `brain.transact()` with the same operations to make it real.
*
* Speculative semantics (documented deltas from the durable paths):
* - `add` without an explicit `vector` produces an entity with an empty
* stub vector speculative entities are not semantically searchable
* (`with()` never invokes the embedder).
* - `remove` cascades on the read side: relations touching the removed
* entity disappear from `related()` without being enumerated.
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* - `relate` deduplicates against the edges visible in this view (overlay
* first, then the record layer), mirroring `brain.relate()`.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* - Brain-level vocabulary/subtype enforcement does not run overlays
* never persist, so there is nothing to protect.
*
* The returned view takes its own pin on this generation; release both
* views independently.
*
* @param ops - The same declarative operations `brain.transact()` accepts.
* @returns A new speculative `Db` over this view.
* @throws EntityNotFoundError when an operation references a missing
* entity, mirroring the durable path's planning errors.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*/
async with(ops: TxOperation<T>[]): Promise<Db<T>> {
this.assertUsable('with')
if (!Array.isArray(ops) || ops.length === 0) {
throw new Error('with(): provide a non-empty array of transaction operations')
}
// Child overlay starts as a copy of this view's overlay (chained with()).
const overlay: SpeculativeOverlay<T> = {
nouns: new Map(this.overlay?.nouns ?? []),
verbs: new Map(this.overlay?.verbs ?? [])
}
// Reads during planning must see "this view + overlay so far".
const speculativeGet = async (id: string): Promise<Entity<T> | null> => {
if (overlay.nouns.has(id)) return overlay.nouns.get(id) ?? null
return this.get(id)
}
for (const op of ops) {
switch (op.op) {
case 'add': {
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
// Reserved-field normalization — mirror of the brain.transact()
// write path: user-settable fields lift to their dedicated field
// (top-level wins), system-managed fields drop, and the entity's
// metadata bag carries ONLY custom fields. Speculative views skip
// the one-shot warnings — committing the same ops through
// `brain.transact()` warns on the real write path.
const { reserved, custom } = splitNounMetadataRecord(
op.metadata as Record<string, unknown> | undefined
)
const confidence =
op.confidence ?? (typeof reserved.confidence === 'number' ? reserved.confidence : undefined)
const weight =
op.weight ?? (typeof reserved.weight === 'number' ? reserved.weight : undefined)
const subtype =
op.subtype ?? (typeof reserved.subtype === 'string' ? reserved.subtype : undefined)
const service =
op.service ?? (typeof reserved.service === 'string' ? reserved.service : undefined)
// Id normalization (8.0) — mirror of the committed transact() add
// path: a natural key coerces to a STABLE UUID (v5), preserving the
// original under ORIGINAL_ID_KEY; a real UUID passes through; no id
// mints a fresh id so the overlay keys match the durable path.
const { id, originalId } = coerceNewEntityId(op.id)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
const now = Date.now()
overlay.nouns.set(id, {
id,
vector: op.vector ?? [],
type: op.type,
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(subtype !== undefined && { subtype }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
data: op.data,
metadata: {
...(custom as object),
...(originalId !== undefined && { [ORIGINAL_ID_KEY]: originalId })
} as T,
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(service !== undefined && { service }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
createdAt: now,
updatedAt: now,
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(confidence !== undefined && { confidence }),
...(weight !== undefined && { weight }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
_rev: 1
})
break
}
case 'update': {
// Id normalization (8.0): resolve a natural key to the canonical UUID
// the write path stored. A real UUID passes through.
const updateId = resolveEntityId(op.id)
const base = await speculativeGet(updateId)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (!base) {
throw new EntityNotFoundError(
updateId,
`with(): entity ${updateId} not found at generation ${this.gen}`
)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
// Same reserved-field normalization as the committed update path.
const { reserved, custom } = splitNounMetadataRecord(
op.metadata as Record<string, unknown> | undefined
)
const confidence =
op.confidence ?? (typeof reserved.confidence === 'number' ? reserved.confidence : undefined)
const weight =
op.weight ?? (typeof reserved.weight === 'number' ? reserved.weight : undefined)
const subtype =
op.subtype ?? (typeof reserved.subtype === 'string' ? reserved.subtype : undefined)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
const mergedMetadata =
op.merge !== false
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
? ({ ...(base.metadata as object), ...custom } as T)
: ((op.metadata !== undefined ? custom : base.metadata) as T)
overlay.nouns.set(updateId, {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
...base,
...(op.type !== undefined && { type: op.type }),
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(subtype !== undefined && { subtype }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
...(op.data !== undefined && { data: op.data }),
...(op.vector !== undefined && { vector: op.vector }),
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(confidence !== undefined && { confidence }),
...(weight !== undefined && { weight }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
metadata: mergedMetadata,
updatedAt: Date.now(),
_rev: (base._rev ?? 1) + 1
})
break
}
case 'remove': {
// Id normalization (8.0): resolve a natural key to the canonical UUID
// the write path stored. A real UUID passes through.
const removeId = resolveEntityId(op.id)
overlay.nouns.set(removeId, null)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
// Tombstone overlay relations touching the removed entity; base
// relations are cascade-filtered at read time in related().
for (const [verbId, relation] of overlay.verbs) {
if (relation && (relation.from === removeId || relation.to === removeId)) {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
overlay.verbs.set(verbId, null)
}
}
break
}
case 'relate': {
// Id normalization (8.0) — mirror of relate(): resolve BOTH endpoints
// to the canonical UUID the write path stored. Real UUIDs pass through.
const relFrom = resolveEntityId(op.from)
const relTo = resolveEntityId(op.to)
const from = await speculativeGet(relFrom)
const to = await speculativeGet(relTo)
if (!from) {
throw new EntityNotFoundError(relFrom, `with(): source entity ${relFrom} not found`)
}
if (!to) {
throw new EntityNotFoundError(relTo, `with(): target entity ${relTo} not found`)
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
// Dedupe against the view (overlay first, then the edges visible
// at this generation via the record layer) — mirror of relate().
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
let duplicate: Relation<T> | undefined
for (const relation of overlay.verbs.values()) {
if (relation && relation.from === relFrom && relation.to === relTo && relation.type === op.type) {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
duplicate = relation
break
}
}
if (!duplicate) {
const existing = await this.related({ from: relFrom, type: op.type })
duplicate = existing.find((relation) => relation.to === relTo)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
if (duplicate) break
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
// Reserved-field normalization — relationship mirror of the add
// op above (and of the committed relate() path).
const { reserved, custom } = splitVerbMetadataRecord(
op.metadata as Record<string, unknown> | undefined
)
const confidence =
op.confidence ?? (typeof reserved.confidence === 'number' ? reserved.confidence : undefined)
const weight =
op.weight ?? (typeof reserved.weight === 'number' ? reserved.weight : undefined)
const subtype =
op.subtype ?? (typeof reserved.subtype === 'string' ? reserved.subtype : undefined)
const service =
op.service ?? (typeof reserved.service === 'string' ? reserved.service : undefined)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
const id = uuidv4()
overlay.verbs.set(id, {
id,
from: relFrom,
to: relTo,
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
type: op.type,
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(subtype !== undefined && { subtype }),
weight: weight ?? 1.0,
...(confidence !== undefined && { confidence }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
...(op.data !== undefined && { data: op.data }),
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
metadata: custom as T,
...(service !== undefined && { service }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
createdAt: Date.now()
})
if (op.bidirectional) {
const reverseId = uuidv4()
overlay.verbs.set(reverseId, {
id: reverseId,
from: relTo,
to: relFrom,
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
type: op.type,
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
...(subtype !== undefined && { subtype }),
weight: weight ?? 1.0,
...(confidence !== undefined && { confidence }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
...(op.data !== undefined && { data: op.data }),
feat(8.0): reserved-field contract — one canonical location, typed prevention, unified read/write Brainy-owned field names (noun/verb, subtype, createdAt, updatedAt, confidence, weight, service, data, createdBy, _rev) now have exactly one home — top level — enforced by three layers driven from a single source of truth, src/types/reservedFields.ts (RESERVED_ENTITY_FIELDS / RESERVED_RELATION_FIELDS, exported): 1. Compile time — AddParams/UpdateParams/RelateParams/UpdateRelationParams metadata (and the transact() ops that extend them) reject a literal reserved key as a TypeScript error while keeping generic T ergonomics (typed bags, untyped brains, index-signature shapes, and a documented exemption for T-declared reserved keys). Pinned by @ts-expect-error type tests run under vitest typecheck mode on every unit run. 2. Write time — the 7.x update() remap is ported to 8.0 and extended to every write path: add/update/relate/updateRelation, their transact() mirrors, and db.with() overlays. User-settable fields lift to their dedicated param (top-level wins when both are supplied — closes the 7.x trap where update({metadata:{confidence}}) silently no-oped), and system-managed fields drop with a one-shot warning naming the right path. A remapped subtype satisfies subtype-pairing enforcement exactly like a top-level one. 3. Read time — every storage combine goes through one canonical hydration helper (hydrateNounWithMetadata / hydrateVerbWithMetadata over splitNoun/VerbMetadataRecord), so reserved fields surface ONLY top-level and entity/relation.metadata carry ONLY custom fields on live reads, batch reads, paginated listings, getRelations by source/target, streamed verbs, and historical asOf() materialization alike. Read-path echoes found and fixed (previously the full stored record — including the verb type key — leaked inside metadata): noun pagination, verb pagination, getVerbsBySource/ByTarget (adjacency + shard fallback), getVerbsBySourceBatch (which also dropped subtype/data), and the filesystem verb stream. getRelations() results now surface confidence/updatedAt top-level via verbsToRelations, updateRelation() no longer erases service/createdBy, relate() persists its top-level confidence/service params, and the dead convertHNSWVerbToGraphVerb echo path is deleted. Import paths (CLI extract, deduplicator, coordinators, neural import) write confidence through the dedicated param instead of the bag. UpdateRelationParams is now exported from the package root. Documented for consumers in docs/concepts/consistency-model.md ("Reserved fields") and RELEASES.md. Regression tests ported from the 7.x fix and extended to the full 8.0 contract (17 runtime tests + 41 type-level assertions); full unit suite 1427/1427, db-mvcc integration 24/24.
2026-06-11 13:12:50 -07:00
metadata: custom as T,
...(service !== undefined && { service }),
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
createdAt: Date.now()
})
}
break
}
case 'unrelate': {
overlay.verbs.set(op.id, null)
break
}
default: {
const exhaustive: never = op
throw new Error(`with(): unknown operation ${JSON.stringify(exhaustive)}`)
}
}
}
this.host.pinGeneration(this.gen)
return new Db<T>({
host: this.host,
generation: this.gen,
timestamp: Date.now(),
overlay
})
}
/**
* @description The ids changed by committed transactions in the interval
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
* `(lowerBound, this.generation]` the diff between an earlier state and this
* view. The lower bound is **exclusive** (contrast `transactionLog`'s inclusive
* window). Single-operation writes between commits are not reflected (documented
* 8.0 history granularity).
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
* The lower bound may be:
* - **a `Db`** an earlier view of the SAME store (its pinned generation);
* - **a generation number** used directly as the exclusive lower bound;
* - **a `Date`** resolved to the generation committed at or before it.
*
* `db.since(prior)` equals `db.since(prior.generation)` by construction.
*
* @param lowerBound - An earlier `Db`, a generation number, or a `Date`.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* @returns Changed entity and relation ids, sorted.
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
* @throws RangeError when the resolved lower bound is newer than this view.
* @throws Error when a `Db` lower bound belongs to a different store.
* @throws GenerationCompactedError when a number/`Date` lower bound is below the
* compaction horizon.
* @example
* const before = brain.now()
* await brain.transact(ops)
* await brain.now().since(before) // ids changed by the transaction
* await brain.now().since(before.generation)
* await brain.now().since(new Date(Date.now() - 3_600_000))
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*/
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
async since(lowerBound: Db<T> | number | Date): Promise<ChangedIds> {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
this.assertUsable('since')
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
let fromGen: number
if (lowerBound instanceof Db) {
if (lowerBound.host.store !== this.host.store) {
throw new Error('since(): both Db values must come from the same Brainy instance')
}
fromGen = lowerBound.gen
} else {
// number = exclusive lower-bound generation; Date resolves via the shared
// asOf resolver (same reachability gate). changedBetween treats fromGen as
// exclusive, so since(db) === since(db.generation).
fromGen = (await this.host.resolveGeneration(lowerBound)).generation
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
if (fromGen > this.gen) {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
throw new RangeError(
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
`since(): lower-bound generation ${fromGen} is newer than this view ` +
`(generation ${this.gen}). Pass an earlier generation/Date, or call newerDb.since(olderDb).`
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
)
}
feat(8.0): temporal range verbs — diff, history, since(gen|Date), asOf{exclusive}, transactionLog window asOf() answers "state AT a point"; these answer "what happened BETWEEN two points" and "one entity's whole history", all on the existing generation records. - Extract resolveAsOfGeneration() as the shared gen|Date→{generation,timestamp} chokepoint (reachability + Date semantics identical everywhere). asOf()'s inclusive path is byte-identical — db-mvcc still 25/25. - asOf(target, { exclusive }) — strict-before (resolved−1, clamped at gen 0, never a RangeError). - db.since(Db | generation | Date) — overload; number/Date resolve via the shared resolver (new DbHost.resolveGeneration); EXCLUSIVE lower bound; since(db) === since(db.generation). Same-store guard via Db.belongsToStore. - brain.diff(a, b) → { added, removed, modified } split by nouns/verbs. EARNS its name: candidate set is ONLY changedBetween(gLow,gHigh), each classified by existence at both endpoints + a key-order-insensitive value compare (new src/db/stableEqual.ts) — a touched-but-reverted / born-and-died id is in no bucket. - brain.history(id, { from, to }) → every distinct version oldest→newest, each value === asOf(version.generation).get(id); null = removal; kind auto-detected (throws on UUID-space collision). New store helper generationsTouching(). - brain.transactionLog({ from, to, limit }) — INCLUSIVE generation/Date window (contrast since's exclusive lower bound); limit applied last. - Compaction policy locked: diff/since THROW GenerationCompactedError below the horizon; history TRUNCATES to it. Types DiffResult/HistoryVersion/EntityHistory exported from the package root. Tests: tests/integration/db-temporal.test.ts (8 proofs incl. diff-earns-its-name, history↔asOf cross-check, the composition proof, granularity, compaction contrast) + tests/unit/db/stableEqual.test.ts (6). docs/guides/snapshots-and-time-travel.md + RELEASES updated. 1477 unit + db-temporal 8 + db-mvcc 25 green.
2026-06-19 13:21:02 -07:00
return this.host.store.changedBetween(fromGen, this.gen)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
// ==========================================================================
// Durability
// ==========================================================================
/**
* @description Snapshot this view to `path` as a self-contained store:
* flush, then hard-link every immutable file into the target
* (Cassandra-style instant and space-shared; cross-device targets fall
* back to byte copies, and the small append-in-place file list is always
* byte-copied). The result opens with `Brainy.load(path)` with the full
* query surface, and later mutations of the source can never alter it
* rewrites swap inodes, the snapshot keeps the old bytes.
*
* `persist()` requires this view to still be the store's **latest**
* generation (snapshotting captures current bytes): pin with `brain.now()`
* and persist before further writes. A view that history has moved past
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* throws {@link GenerationConflictError}; speculative overlays throw
* {@link SpeculativeOverlayError} (commit them with `brain.transact()`
* first).
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*
* @param path - Absolute directory for the snapshot (created; must be
* empty or absent).
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* @throws GenerationConflictError when this view is no longer the latest
* generation.
* @throws SpeculativeOverlayError when this view is a speculative overlay.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*/
async persist(path: string): Promise<void> {
this.assertUsable('persist')
if (this.overlay) {
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
throw new SpeculativeOverlayError('persist', this.gen)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
await this.host.persistPinned(path, this.gen)
}
// ==========================================================================
// Lifecycle
// ==========================================================================
/**
* @description Release this view's pin (store + versioned index
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* providers) and free its at-generation index materialization, if one was
* built. Idempotent. After release, every read throws. Views from
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* `Brainy.load()` / `asOf(path)` also close their underlying read-only
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* brain here. A `FinalizationRegistry` backstop releases leaked pins (and
* materializations) when a `Db` is garbage-collected, but explicit
* release is what makes `compactHistory()` deterministic prefer it.
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*/
async release(): Promise<void> {
if (this.isReleased) return
this.isReleased = true
this.host.unregisterDbFromFinalization(this)
this.host.releaseGeneration(this.gen)
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
if (this.materializedHandle) {
// A failed materialization already surfaced to the query that
// triggered it; release() must still succeed.
const handle = await this.materializedHandle.catch(() => null)
if (handle) await handle.close()
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (this.closeOnRelease) {
await this.closeOnRelease()
}
}
// ==========================================================================
// Internals
// ==========================================================================
/** Whether any transaction committed past the pinned generation. */
private isHistorical(): boolean {
return this.host.store.hasCommittedAfter(this.gen)
}
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/**
* Build (once) and cache the at-generation index materialization. The
* first index-accelerated query at a historical pin pays the O(n at G)
* build; every later one reuses the cached handle until release(). The
* GC-backstop registration is refreshed so a leaked `Db` also closes its
* materialized reader (whose flush timers would otherwise outlive it).
*/
private materialize(): Promise<HistoricalQueryHandle<T>> {
if (!this.materializedHandle) {
this.materializedHandle = this.host.materializeAt(this.gen).then((handle) => {
this.host.unregisterDbFromFinalization(this)
this.host.registerDbForFinalization(this, this.gen, async () => {
await handle.close()
if (this.closeOnRelease) await this.closeOnRelease()
})
return handle
})
// A failed build must not poison the cache — the next query retries.
this.materializedHandle = this.materializedHandle.catch((err) => {
this.materializedHandle = undefined
throw err
})
}
return this.materializedHandle
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/** Throw on use-after-release. */
private assertUsable(method: string): void {
if (this.isReleased) {
throw new Error(
`Cannot call ${method}() on a released Db (generation ${this.gen}). ` +
`Pin a fresh view with brain.now() or brain.asOf().`
)
}
}
/**
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
* Reject find() dimensions that require index machinery a speculative
* overlay cannot answer (overlay entities carry no embeddings and are in
* no index see {@link SpeculativeOverlayError}).
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
*/
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
private assertOverlayCompatibleFind(params: FindParams<T>): void {
for (const [capability, present] of indexOnlyFindDimensions(params)) {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (present) {
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
throw new SpeculativeOverlayError(capability, this.gen)
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
}
}
}
}
feat(8.0): full query surface at historical generations via ephemeral index materialization Historical Db values (now()/asOf() pins that history has moved past) now serve the COMPLETE query surface - vector/hybrid search, graph traversal, cursor pagination, and aggregation - by materializing ephemeral in-memory indexes over the exact at-generation record set. The historical-query throw is gone; NotYetSupportedAtHistoricalGenerationError is deleted. Materializer (Brainy.materializeAtGeneration): - Copies the at-G record set (live bytes for ids untouched since the pin, immutable before-images otherwise) into a fresh MemoryStorage; a final reconciliation pass under the commit mutex makes the copy exact even when transactions commit mid-build. - Opens a read-only Brainy over the copy: init rebuilds the metadata and graph-adjacency indexes from the records; the vector index is built by inserting every at-G vector (the at-G HNSW graph never existed on disk, so there is nothing to restore). Host embedder and aggregate definitions are shared - no second model load, aggregates backfill at-G values. - Cost is the documented contract: O(n at G) time and memory, ONCE per Db (handle cached; freed by release(), with a FinalizationRegistry backstop that also closes leaked readers). A native VersionedIndexProvider serves the same reads from retained segments with no rebuild. Db routing (src/db/db.ts): metadata-level find()/related() keep the free record path; index-only dimensions (query/vector/near/connected/cursor/ aggregate/includeRelations/non-metadata modes) route to the cached materialization; unsupported where-operators on the record path re-route there too instead of erroring. Speculative with() overlays keep the one honest boundary - SpeculativeOverlayError (overlay entities carry no embeddings, so index reads over them would be silently incomplete); metadata find()/get()/filter related() work on overlays. UpdateParams.vector contract now honored: an explicit pre-computed vector applies directly (with dimension validation) in update() and transact update ops, re-indexing HNSW - previously it was silently ignored unless data also changed. GraphAdjacencyIndex: adjacency now derives from the two verb-id LSM trees filtered through the live-verb tombstone set (entity->entity edge trees deleted - they carried no verb ids, so removeVerb could never tombstone them and traversal served stale neighbors forever). Neighbor reads batch- load live verbs via the unified cache; addVerb seeds the cache. Proofs (tests/integration/db-mvcc.test.ts, 24 green): historical vector search finds old vector placement including since-deleted entities; historical graph traversal walks the old wiring after a rewire; historical aggregation computes at-G group values; asOf() pins get the same surface; the materialization builds once per Db and release() closes the ephemeral reader (it refuses reads afterwards); overlays throw the documented error. ADR-001 updated to the no-throws historical model.
2026-06-11 08:12:11 -07:00
/**
* @description The find() dimensions that only index machinery can answer
* (everything beyond metadata filters + ordering + offset/limit windows),
* as `[capability, present]` pairs for the given params.
*/
function indexOnlyFindDimensions(params: FindParams): Array<[string, boolean]> {
return [
['semantic query', params.query !== undefined],
['vector search', params.vector !== undefined],
['proximity search (near)', params.near !== undefined],
['graph traversal (connected)', params.connected !== undefined],
['cursor pagination', params.cursor !== undefined],
['aggregation', params.aggregate !== undefined],
['relation expansion (includeRelations)', params.includeRelations === true],
[
`search mode '${params.mode ?? params.searchMode}'`,
(params.mode !== undefined && params.mode !== 'metadata' && params.mode !== 'auto') ||
(params.searchMode !== undefined && params.searchMode !== 'auto')
]
]
}
/**
* @description Whether the find() params include any index-only dimension
* the routing predicate that decides between the record path (cheap,
* metadata-only) and the at-generation index materialization at historical
* generations.
*/
function findRequiresIndexes(params: FindParams): boolean {
return indexOnlyFindDimensions(params).some(([, present]) => present)
}
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
/**
* @description Build a `Result` row from a record/overlay-resolved entity,
* mirroring the live metadata-only find path (flattened fields, score `1.0`).
*/
function resultFromEntity<T>(entity: Entity<T>): Result<T> {
return {
id: entity.id,
score: 1.0,
type: entity.type,
subtype: entity.subtype,
metadata: entity.metadata,
data: entity.data,
confidence: entity.confidence,
weight: entity.weight,
_rev: entity._rev,
entity
}
}
/**
* @description In-place stable sort of merged results by a find() `orderBy`
* field, mirroring live ordering semantics (`asc` default, undefined values
* last in both directions).
*/
function sortResultsBy<T>(results: Result<T>[], orderBy: string, order: 'asc' | 'desc'): void {
const direction = order === 'desc' ? -1 : 1
results.sort((a, b) => {
const va = resolveEntityField(a.entity as Entity, orderBy)
const vb = resolveEntityField(b.entity as Entity, orderBy)
if (va === undefined && vb === undefined) return 0
if (va === undefined) return 1
if (vb === undefined) return -1
if (typeof va === 'number' && typeof vb === 'number') return (va - vb) * direction
const sa = String(va)
const sb = String(vb)
return (sa < sb ? -1 : sa > sb ? 1 : 0) * direction
})
}
/**
* @description Filter one relation against `RelatedParams`, mirroring
* the storage-level filter `brain.related()` builds (`from` sourceId,
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
* `to` targetId, type/subtype set membership, service equality).
*/
function relationMatchesParams<T>(relation: Relation<T>, params: RelatedParams): boolean {
feat(8.0): generational MVCC storage + Datomic-style Db API (now/transact/asOf/with/persist) One mechanism replaces the COW and versioning subsystems: immutable generation-stamped records behind a Datomic-style database value. Record layer (src/db/generationStore.ts): - Monotonic generation counter in _system/generation.json; bumped once per transact() commit and once per single-operation write (storage hook), so brain.generation() is always a meaningful watermark. - Commit protocol: stage before-images + tx.json delta -> fsync -> execute batch via TransactionManager -> atomic tmp+rename of _system/manifest.json (the rename IS the commit point) -> append _system/tx-log.jsonl. - Crash recovery on open rolls uncommitted generations back byte-identically and forces an index rebuild; refcounted pins gate compactHistory(), which records a horizon (asOf below it throws GenerationCompactedError). Db API: - Db: get/find/search/related pinned at a generation, with() speculative overlays, since() diffs, persist() hard-link-farm snapshots, timestamp, generation, release() + FinalizationRegistry backstop. - Brainy: now() O(1) pin, transact(ops, {meta, ifAtGeneration}) atomic batch (GenerationConflictError CAS), asOf(generation|Date|path), restore(path, {confirm}), compactHistory(), generation(), static open() + load(). - get()/metadata find()/related() are fully correct at any reachable pinned generation; index-accelerated queries at historical generations throw NotYetSupportedAtHistoricalGenerationError - never silently-wrong results. - VersionedIndexProvider (generation/isGenerationVisible/pin/release) in plugin.ts: feature-detected, balanced pin/release in lockstep with Db lifecycle, post-commit applier + replay-gap model documented. Storage primitives (BaseStorage + filesystem/memory adapters): raw-object read/write/list/remove, fsync barrier, noun/verb raw before-image capture, tx-log append, snapshotToDirectory (hard-link farm; byte-copy fallback and append-in-place exceptions), restoreFromDirectory + derived-state reload. Proof suite (tests/integration/db-mvcc.test.ts + tests/unit/db/): isolation across 200 mutations, batch atomicity under injected execution failure, ifAtGeneration CAS, snapshot immunity to source mutation, compaction safety under pins, with() overlay isolation, generation monotonicity across reopen, crash consistency through the real recovery path, and versioned provider pin/release balance. Design record in docs/ADR-001-generational-mvcc.md.
2026-06-10 14:14:07 -07:00
if (params.from !== undefined && relation.from !== params.from) return false
if (params.to !== undefined && relation.to !== params.to) return false
if (params.type !== undefined) {
const types = Array.isArray(params.type) ? params.type : [params.type]
if (!types.includes(relation.type)) return false
}
if (params.subtype !== undefined) {
const subtypes = Array.isArray(params.subtype) ? params.subtype : [params.subtype]
if (relation.subtype === undefined || !subtypes.includes(relation.subtype)) return false
}
if (params.service !== undefined && relation.service !== params.service) return false
return true
}