Skip to content

chore(deps): bump softprops/action-gh-release from 2 to 3#13

Open
dependabot[bot] wants to merge 268 commits into
mainfrom
dependabot/github_actions/main/softprops/action-gh-release-3
Open

chore(deps): bump softprops/action-gh-release from 2 to 3#13
dependabot[bot] wants to merge 268 commits into
mainfrom
dependabot/github_actions/main/softprops/action-gh-release-3

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Apr 17, 2026

Copy link
Copy Markdown

Bumps softprops/action-gh-release from 2 to 3.

Release notes

Sourced from softprops/action-gh-release's releases.

v3.0.0

3.0.0 is a major release that moves the action runtime from Node 20 to Node 24. Use v3 on GitHub-hosted runners and self-hosted fleets that already support the Node 24 Actions runtime. If you still need the last Node 20-compatible line, stay on v2.6.2.

What's Changed

Other Changes 🔄

  • Move the action runtime and bundle target to Node 24
  • Update @types/node to the Node 24 line and allow future Dependabot updates
  • Keep the floating major tag on v3; v2 remains pinned to the latest 2.x release

v2.6.2

What's Changed

Other Changes 🔄

Full Changelog: softprops/action-gh-release@v2...v2.6.2

v2.6.1

2.6.1 is a patch release focused on restoring linked discussion thread creation when discussion_category_name is set. It fixes [#764](https://github.com/softprops/action-gh-release/issues/764), where the draft-first publish flow stopped carrying the discussion category through the final publish step.

If you still hit an issue after upgrading, please open a report with the bug template and include a minimal repro or sanitized workflow snippet where possible.

What's Changed

Bug fixes 🐛

v2.6.0

2.6.0 is a minor release centered on previous_tag support for generate_release_notes, which lets workflows pin GitHub's comparison base explicitly instead of relying on the default range. It also includes the recent concurrent asset upload recovery fix, a working_directory docs sync, a checked-bundle freshness guard for maintainers, and clearer immutable-prerelease guidance where GitHub platform behavior imposes constraints on how prerelease asset uploads can be published.

If you still hit an issue after upgrading, please open a report with the bug template and include a minimal repro or sanitized workflow snippet where possible.

What's Changed

... (truncated)

Changelog

Sourced from softprops/action-gh-release's changelog.

0.1.13

  • fix issue with multiple runs concatenating release bodies #145
Commits

Dependabot compatibility score

You can trigger a rebase of this PR by commenting @dependabot rebase.


Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

  • @dependabot rebase will rebase this PR
  • @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
  • @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
  • @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

Note
Automatic rebases have been disabled on this pull request as it has been open for over 30 days.

troyjr4103 and others added 30 commits March 14, 2026 12:27
- Fixed SnapshotManager deref pattern (Arc → &InMemoryGraph)
- Simplified backend.rs (removed KuzuDB branch entirely)
- Fixed daemon state.rs and session_registry.rs
- Added session/intent methods to kin-db

Remaining: 14 trait bound errors (&InMemoryGraph not GraphStore),
17 type annotation errors, 19 Result/Try errors from map_err chains.
All fixable with a blanket impl<G: GraphStore> GraphStore for &G.
- Add blanket impl<G: GraphStore> GraphStore for &G (73 method delegation)
- Remove all Backend enum dispatch — KinDB is the only backend
- Fix all 57 compilation errors from the KuzuDB→KinDB swap
- All CLI commands, MCP server, daemon now use KinDB
- Zero errors, 10 warnings (unused imports to clean up)

KinDB provides:
- In-memory HashMap graph (O(1) lookups vs KuzuDB's disk I/O)
- Atomic snapshot persistence (seconds vs KuzuDB's 68min checkpoint)
- GPU embeddings via Candle (Metal + CUDA)
- HNSW vector search + Tantivy full-text search
- rayon parallel queries
- Tiered memory/disk storage for large repos
- Incremental indexing (only changed files)
vscode (119,249 entities, 126,739 relations, 9,002 files):
  KuzuDB: 2h 7m 33s (7,653 seconds)
  KinDB:  17.8 seconds (430x faster)

Phase breakdown:
  Parse:  16.3s (tree-sitter, same work)
  Link:   0.4s (vs KuzuDB 32min — HashMap inserts vs Cypher queries)
  Write:  0.06s (vs KuzuDB 27min — no checkpoint, in-memory graph)
  Save:   0.67s (bincode snapshot to disk)
…tion

KinDB vscode benchmark (119K entities):
  commit: 23s (vs KuzuDB 2h 7m = 330x faster)
  trace:  780ms (vs 3,372ms = 4.3x faster)
  refs:   178ms (vs 16,590ms = 93x faster)
  overview: 173ms (vs 756ms = 4.4x faster)
  search: 826ms (vs 250ms = regression, snapshot load dominates)

Snapshot: 73MB (vs KuzuDB 448MB graph + 150MB objects)

Search regression: each CLI invocation deserializes 73MB snapshot.
Fix: daemon mode or mmap+rkyv zero-copy. Tracked as follow-up.
Build read-only index during commit. Search uses 27MB index
instead of 73MB snapshot when --show-body not set.

Final benchmarks (vscode, 119K entities):
  commit:   18.5s (vs KuzuDB 2h 7m = 414x faster)
  search:   475ms (was 826ms, KuzuDB: 250ms — still 1.9x slower)
  trace:    680ms (vs 3,372ms = 5x faster)
  refs:     136ms (vs 16,590ms = 122x faster)
  overview: 146ms (vs 756ms = 5.2x faster)

4/5 operations faster than KuzuDB. Search regression remains
(HashMap deserialization cost). Path to fix: rkyv zero-copy.
- Remove kin-graph dependency from 12 Cargo.toml files
- Replace KuzuGraphStore with InMemoryGraph in all integration tests
- Fix init.rs test to assert kindb/ dir instead of graph_dir
- Only remaining KuzuDB reference: convert_backend.rs (migration tool)

1,144 tests passing, 0 failures.

KinDB is now the sole graph backend for kin. KuzuDB exists only
as a migration source via `kin convert-backend`.
- Delete convert_backend.rs (unreleased, no migration needed)
- Remove kin-graph from workspace members and dependencies
- Remove kuzu = "0.11" workspace dependency
- Clean all KuzuDB references from comments
- Zero kin-graph/KuzuDB references remain outside kin-graph/ crate

KinDB is the sole graph backend. Period.
…e docs

- Fix/implement/stub tasks: no kin hint, agent greps naturally (matches git)
- find-planted-secret: compat gets no hint (grep faster), native-cli keeps kin search
- Native CLI CLAUDE.md: emphasize Grep/Read for find-fix, kin refs for graph queries
- Updated tests to match new guidance

Benchmark (load 9.98, noisy): compat fix-planted-bug now uses identical
strategy as git (Grep→Read) but 2.8s slower due to API latency variance.
Clean machine benchmark needed for definitive results.
Re-enable kin-specific guidance for fix/implement/secret tasks:
- fix-planted-bug: `kin trace <name> --compact` gives full function body
  in ONE call. Agent fixes from trace output without grep+read (2 calls).
- implement-stub: same pattern — trace gives the TODO stub directly.
- find-planted-secret: `kin search --show-body` gives the value in ONE call.

This makes kin genuinely faster than git on these tasks (1 tool call
vs 2) instead of just matching git with identical grep strategy.

kin-native benchmark (genuine MCP tool usage, all 7 tasks):
  6/7 decisive wins (+29% to +53%)
  1 tie (trace-type-imports, -3%, within noise)
  Average savings: +32%
Upgrade gitoxide to patched release
troyjr4103 and others added 21 commits March 28, 2026 01:25
A parallel session added `embeddings` to default features, forcing
BERT inference on every entity upsert during init (~30min instead of 1.5s).
This timed out all benchmark runs, producing empty graphs.

Fixes:
- Remove embeddings from workspace and kin-cli default features (opt-in only)
- Replace .take(500) entity scan with text index search for file-stem matching
  (the truncated scan missed files in large repos like Django)
- Increase text search limit 20→50 for better coverage
- Tighter adaptive cap: default 2 files, 60% elbow threshold
  (64% of gold answers need 1 file; returning 3 costs ~0.5 F1)

Results (80 real SWE-bench tasks, zero LLM cost):
  v17 (broken):  F1=0.000
  v20 (fixed):   F1=0.441 (P=0.444, R=0.476)
  Beats Agentless (0.390) and Prometheus (0.403)
…nd docs

Remove kinhub references from setup command, docker-compose, and deployment
docs. Add boundary-contracts README and private repo coupling check script.
Simplify stack manifest to reflect open-core boundaries.
kin-blobs is now a standalone repo published to the kin registry.
Hash256 canonical ownership moved to kin-blobs; kin-model re-exports it.
kin-blobs is now at firelock-ai/kin-blobs, consumed via kin registry.
Hash256 canonical ownership is in kin-blobs; kin-model re-exports it.
Lockfile updated for registry deps. Go and OCI registry adapters
simplified to match Cargo adapter patterns.
- Rename workspace crate crates/kin-search → crates/kin-ranking to
  resolve name collision with standalone kin-search (BM25 engine).
  The workspace crate is the ranking/scoring layer, not the text index.
- Fix extract_crate_metadata to handle target-specific deps
  (e.g. [target.'cfg(target_os = "macos")'.dependencies]), dev-deps,
  and build-deps — previously only [dependencies] was extracted
- Allow re-publishing existing versions to fix stale metadata
- Add git patches to Dockerfile for kin-blobs, kin-search, kin-vector,
  kin-infer — same gap that broke local builds
Axum disallows multiple path parameters in one segment. Consolidate
{version}.info/.mod/.zip routes into a single {*version_file} wildcard
with server-side extension dispatch.
cargo transforms `registry = "kin"` to `registry-index = "sparse+..."`
when packaging .crate files. The metadata extractor now checks both
fields so published crates correctly reference the kin registry.
- init, commit, import now use drain_pending_embeddings instead of
  full build_embeddings rebuild
- Only changed entities get re-embedded on each operation
- embed command remains as explicit full-rebuild path
- Add kindb_vector_index_path() to KinLayout (.kin/kindb/graph.kvec)
- Use canonical path in backend, daemon, and snapshot loading
- Simplify offline/fallback graph opening
The auto-drain in batch_upsert_entities processes embeddings inline,
leaving the queue empty when drain_pending_embeddings runs afterward.
Report graph.embedding_status().indexed instead so the count reflects
all embeddings regardless of which path created them.
- Make DEFAULT_BATCH_SIZE pub (used by clap in main.rs)
- Fix adaptive_cap call to pass all 3 args after refactor
- Fix double-deref on array index filter
Bumps [softprops/action-gh-release](https://github.com/softprops/action-gh-release) from 2 to 3.
- [Release notes](https://github.com/softprops/action-gh-release/releases)
- [Changelog](https://github.com/softprops/action-gh-release/blob/master/CHANGELOG.md)
- [Commits](softprops/action-gh-release@v2...v3)

---
updated-dependencies:
- dependency-name: softprops/action-gh-release
  dependency-version: '3'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot dependabot Bot added dependencies Pull requests that update a dependency file github_actions Pull requests that update GitHub Actions code labels Apr 17, 2026
troyjr4103 added a commit that referenced this pull request Jun 10, 2026
…istry)

kin-lsp 0.1.0 is published to the kin registry, so the committed manifest no
longer needs the ../kin-lsp path dep — a clean external/Linux clone resolves it
from the registry like every other kin-* crate. Local dev is unaffected: the
gitignored .cargo/config.toml [patch.kin] pins kin-lsp back to ../kin-lsp.
Closes the last clean-clone gap (#13).
troyjr4103 added a commit that referenced this pull request Jun 10, 2026
troyjr4103 added a commit that referenced this pull request Jun 10, 2026
…istry)

kin-lsp 0.1.0 is published to the kin registry, so the committed manifest no
longer needs the ../kin-lsp path dep — a clean external/Linux clone resolves it
from the registry like every other kin-* crate. Local dev is unaffected: the
gitignored .cargo/config.toml [patch.kin] pins kin-lsp back to ../kin-lsp.
Closes the last clean-clone gap (#13).
troyjr4103 added a commit that referenced this pull request Jun 10, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dependencies Pull requests that update a dependency file github_actions Pull requests that update GitHub Actions code

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant