chore: release v0.7.0

CHANGELOG.md

@@ -1,5 +1,142 @@
 # Changelog
 
+## v0.7.0 - 2026-10-03
+
+### Added
+
+- *(query)* [**breaking**] procedure catalog with typed signatures
+- *(cdc)* [**breaking**] explicit retention policy and acknowledged progress per consumer
+- *(vector)* remove nodes and reuse their slots beside live search
+- schema claims, event-driven background work
+- journal index writes, speed up the Raft path
+- *(server)* add a healthcheck command backed by a real /ready
+- *(query)* let the caller bound the wait for a building vector index
+- *(server)* [**breaking**] drop seqno-bounded incremental backups
+- *(embed)* run a TTL reap pass on demand
+- *(modality)* insert vectors under the shared index lock
+- *(storage)* report local commits on the applied feed
+- *(vector)* insert through a shared borrow, concurrently with search
+- *(vector)* track whether each node is live
+
+### Documentation
+
+- *(server)* drop a pointer to an internal log from a test header
+- *(server)* drop internal ids and document paths
+- drop internal task and decision ids from four modules
+- state the rules in source comments instead of pointing at a local file
+- *(cluster)* drop internal ids and document paths
+- state design reasons instead of citing other codebases
+- drop internal ids from core, modality, site theme and examples
+- *(core,embed)* replace leftover document stand-ins in comments
+- *(core)* drop internal ids and document paths
+- *(embed)* resolve the RestoreError links in Database::restore
+- *(embed)* drop internal ids and document paths
+- *(modality)* drop internal ids and document paths
+- *(query)* drop internal ids and document paths
+- *(raft)* drop internal ids and document paths
+- drop internal ids from search, replicate, swarm, wire and timeseries
+- *(replicate)* stop linking a private constant from public docs
+- *(storage)* state the durability invariants instead of their ids
+- *(storage)* drop internal ids and document paths
+- *(vector)* drop internal ids and document paths
+
+### Fixed
+
+- *(server)* explain the plan ExecuteCypher runs
+- *(server)* report an auto-commit write conflict as retryable
+- *(cdc)* refuse a resume past purged log instead of skipping
+- *(restore)* [**breaking**] keep every NodeId, refuse issued ones, restore schema and indexes
+- *(build)* keep aws-lc out of the server binary
+- *(raft)* keep snapshots in a file, not in memory or the store
+- *(server)* refuse MiB sizes that do not fit instead of clamping
+- *(server)* refuse unknown schema enum values instead of guessing
+- *(raft)* wait for leadership transfers and checkpoints to finish
+- *(server)* bind the gRPC port before opening storage
+- *(server)* keep the hop count of a forwarded request
+- *(server,s3)* stop deleting chunks another blob still reads
+- *(server)* report an unusable storage topology instead of panicking
+- *(server)* roll back idle interactive transactions on a timer
+- *(raft)* bound the linearizable lease read by the fence timeout
+- *(cdc)* keep a slow reader registered; configure stream pacing
+- *(cdc)* stream applied Raft log entries from the log's own files
+- *(vector)* offline an index whose build cannot start
+- *(replicate)* repair partitions at an exact raft position
+- *(raft)* build snapshots from a capture at their log id
+- *(core)* draw proposal ids from a fresh range per process
+- *(query)* keep undeclared properties where reads find them
+- *(query)* match and write every label of a node pattern
+- *(query)* find an unlabelled temporal traversal target
+- *(query)* bind the TRANSFER EDGES edge variable during analysis
+- *(query)* say why a temporal write is refused, without internal ids
+- *(vector)* maintain replica indexes from applied entries
+- *(vector)* build indexes from a tap of applied writes
+- *(storage)* rebuild from what the checkpoint trees hold
+- *(storage)* survive a crash inside a partition rebuild
+- *(storage)* replay the embedded journal by per-partition coverage
+- *(deps)* require the lsm-tree release the code needs
+- *(query)* finish a vector build under writes that never pause
+- *(raft)* settle membership and shutdown under load
+- *(query)* filter anonymous edges on inline properties
+- *(query)* treat Windows and Unix source paths as one call site
+- *(raft)* honour the limiter bypass for every write concern
+- *(raft)* publish the applied index when a snapshot is installed
+- *(raft)* stop without waiting on a group that is gone
+- *(raft)* wait out snapshot work on shutdown
+- *(storage)* [**breaking**] name the upgrade path from 0.6 stores
+- *(raft)* resume and purge the raft log by per-tree apply coverage
+- *(storage)* cover columnar tables in recovery and in snapshots
+- *(replicate)* keep heartbeats visible while their write is in flight
+- keep heartbeats of a failed flush, lsm-tree 5.11.10
+- *(storage)* enforce version conditions on unwritten keys
+- *(storage)* skip delete-pending files in retention stats on Windows
+- *(storage)* keep the raft coverage reset tombstone below its base
+- *(vector)* hand the entry point to an updated node, seed reorder at it
+- *(vector)* keep node metadata at stable addresses
+- *(vector)* recompute neighbour edits after a lost CAS
+- *(vector)* publish layer-0 neighbour lists whole
+- *(vector)* build large insert_batch in rounds
+- *(vector)* publish upper-layer neighbour lists whole
+- *(vector)* keep a reinserted HNSW node reachable
+
+### Performance
+
+- *(storage)* measure the apply pauses of copies and installs
+- *(storage)* measure what apply coverage costs
+- *(vector)* keep publication counters on their own cache lines
+
+### Refactored
+
+- *(query)* pair the vector registry with its build engine
+- *(vector)* shard the id map for concurrent inserts
+- *(vector)* create the node stores through a shared borrow
+- *(vector)* keep node codes in the stable store
+- *(vector)* keep layer-0 nodes at stable addresses
+- *(vector)* keep only RaBitQ codes in the code block
+
+### Testing
+
+- *(server)* CALL listing and refusal through ExecuteCypher
+- *(server)* cover version conditions through the public commit
+- *(server)* name schema proto enums instead of their integers
+- *(storage,raft)* cut power at every sync and write, and upgrade a v0.6.0 store
+- reserve test ports across processes and catch failed starts
+- *(integration)* name tests by behaviour, not by ids
+- make the suites pass on any target dir and as root
+- *(embed)* wait for a node to reach the HNSW graph
+- *(embed)* allow expect in the traversal and label tests
+- *(embed)* find checkpoint-shared tables by name, not inode
+- *(raft)* cover log retention, failed rebuilds and interrupted installs
+- time the watermark advance from the read it unblocks
+- *(raft)* poll for snapshot builds instead of sleeping on them
+- *(raft)* pin the snapshot install's ordering guarantees
+- *(raft)* cut open replication streams in the partition nemesis
+- *(vector)* wait for the build queue instead of sleeping
+- *(vector)* size the layer-0 replace race for miri
+
+### Bench
+
+- *(vector)* search during concurrent inserts
+
 ## v0.6.0 - 2026-09-23
 
 ### Added

Cargo.toml

@@ -31,7 +31,7 @@ exclude = [
 ]
 
 [workspace.package]
-version = "0.6.0"
+version = "0.7.0"
 edition = "2021"
 rust-version = "1.90"
 license = "AGPL-3.0-only"

crates/coordinode-bench/CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All notable changes to this crate are documented in this file.
 
+## v0.7.0 - 2026-10-03
+
+### Documentation
+
+- *(cluster)* drop internal ids and document paths
+
 ## 0.5.2 - 2026-08-30
 
 #### Fixed

crates/coordinode-cluster/CHANGELOG.md

@@ -2,6 +2,16 @@
 
 All notable changes to this crate are documented in this file.
 
+## v0.7.0 - 2026-10-03
+
+### Added
+
+- schema claims, event-driven background work
+
+### Documentation
+
+- *(cluster)* drop internal ids and document paths
+
 ## v0.6.0 - 2026-09-23
 
 ### Documentation

crates/coordinode-core/CHANGELOG.md

@@ -2,6 +2,25 @@
 
 All notable changes to this crate are documented in this file.
 
+## v0.7.0 - 2026-10-03
+
+### Added
+
+- schema claims, event-driven background work
+- journal index writes, speed up the Raft path
+
+### Documentation
+
+- state design reasons instead of citing other codebases
+- drop internal ids from core, modality, site theme and examples
+- *(core,embed)* replace leftover document stand-ins in comments
+- *(core)* drop internal ids and document paths
+
+### Fixed
+
+- *(query)* keep undeclared properties where reads find them
+- *(core)* draw proposal ids from a fresh range per process
+
 ## v0.6.0 - 2026-09-23
 
 ### Added