Skip to content

Actions: photostructure/node-sqlite

Actions

Build & Release

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
build.yml
258 workflow runs
258 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Build - fix(alpine): add --runInBand workaround for N-API musl crash Workaround for upstream issue: N-API Napi::Reference destructors calling napi_delete_reference during Jest worker termination on Alpine/musl corrupt V8's JIT pages, causing SIGSEGV/SIGABRT. Changes: - Add --runInBand to Alpine CI tests (prevents parallel worker termination) - Remove explicit Reset() in ValueStorage::Remove() (reduces crash frequency) - Update P02 documentation with accurate analysis The previous "fix" was never commi... Build & Release #259: Commit 68b1124 pushed by mceachen
21m 12s main
main
21m 12s
Build - chore(tests): sync node Build & Release #258: Commit e738b8c pushed by mceachen
23m 11s main
main
23m 11s
Build - chore(deps-dev): bump the development group across 1 directory with 4 updates Build & Release #257: Pull request #94 opened by dependabot bot
27m 27s dependabot/npm_and_yarn/development-302187baf9
dependabot/npm_and_yarn/development-302187baf9
27m 27s
Build - fix(deps): update eslint-plugin-regexp and prettier to latest versions Build & Release #256: Commit 804d950 pushed by mceachen
30m 1s main
main
30m 1s
Build - chore(deps-dev): bump the development group with 6 updates Build & Release #255: Pull request #93 opened by dependabot bot
25m 19s dependabot/npm_and_yarn/development-638995fa59
dependabot/npm_and_yarn/development-638995fa59
25m 19s
Build - test(ci): add comment about Alpine fix in workflow Document that parallel test execution now works on Alpine after the proper fix (commit 0691ae5) removed database_ref_ from StatementSync. Build & Release #254: Commit 75f1cd2 pushed by mceachen
26m 12s main
main
26m 12s
Build - fix(deps): update tar package to version 7.5.7 Build & Release #253: Commit 8bb8726 pushed by mceachen
28m 30s main
main
28m 30s
Build - revert(Session): remove database_ref_ to fix Alpine/musl crashes Revert Session class to the simple design from b5835fa (Dec 2025), removing the Napi::ObjectReference database_ref_ that was added in commit a151cb6 and subsequent "fixes" (fb283df through 611d330). Root cause: N-API reference manipulation during GC finalization corrupts V8's JIT page allocations on Alpine/musl. The database_ref_ field caused napi_delete_reference to be called from Session destructors during GC, which is documente... Build & Release #252: Commit 4da0638 pushed by mceachen
24m 19s main
main
24m 19s
Build - fix(Session): skip Reset() in destructor to prevent V8 JIT corruption On Alpine/musl, calling Napi::ObjectReference::Reset() during GC finalization corrupts V8's JIT page allocations, causing SIGTRAP/SIGSEGV. The ObjectReference destructor is designed to be GC-safe, but explicitly calling Reset() during GC is not. This fix adds an in_destructor_ flag to Session so that Delete() skips the explicit Reset() call when called from the destructor, letting the ObjectReference destructor handle cleanup... Build & Release #251: Commit 611d330 pushed by mceachen
22m 58s main
main
22m 58s
Build - fix(Session): prevent V8 JIT corruption during cleanup on Alpine Hold strong references to Session JS objects during DeleteAllSessions() cleanup. This prevents GC from destroying Sessions while we iterate, avoiding V8 JIT page corruption that manifested as SIGSEGV on Alpine/musl. The bug: When Sessions were GC'd during Jest cleanup, their ObjectReference destructors corrupted V8's JIT allocations. This only happened on Alpine because musl has different GC timing during process exit. The fix: T... Build & Release #250: Commit 413c93f pushed by mceachen
27m 2s main
main
27m 2s
Build - docs: update P02 with correct fix explanation for dangling pointers Build & Release #249: Commit 5d1edcb pushed by mceachen
1m 0s main
main
1m 0s
Build - fix(Session): prevent dangling pointers in DeleteAllSessions Remove database_ref_.Reset() calls during DeleteAllSessions() iteration. The issue: Reset() can trigger GC, which may finalize Session JS objects that are still in our iteration list, creating dangling pointers → SIGSEGV. Why this is safe: - Our iteration only uses pure C/C++ ops (no Napi calls that trigger GC) - When Sessions are later GC'd, Delete() returns early (session_ is nullptr) - Napi::ObjectReference destructor handles re... Build & Release #248: Commit e1a3fcb pushed by mceachen
28m 24s main
main
28m 24s
Build - docs: document Bug 2d double-free fix in DeleteAllSessions Build & Release #247: Commit 0157393 pushed by mceachen
27m 51s main
main
27m 51s
Build - fix(docs): update investigation report on flaky native crashes with multiple root causes and fixes Build & Release #246: Commit 1c8215c pushed by mceachen
32m 7s main
main
32m 7s
Build - fix(Session): avoid mutex deadlock in DeleteAllSessions Calling database_ref_.Reset() can trigger GC, which can finalize other Session objects. Those Session destructors call Delete() -> RemoveSession() which tries to lock sessions_mutex_. Since std::mutex is not recursive, holding the lock during Reset() causes undefined behavior (SIGSEGV on Alpine/musl). Fix: Release the mutex before iterating. Since sessions_ is cleared first, any RemoveSession() calls during GC become no-ops (erasing from e... Build & Release #245: Commit dadbb86 pushed by mceachen
22m 45s main
main
22m 45s
Build - fix(Session): release database_ref_ in DeleteAllSessions When DeleteAllSessions() sets session->session_ = nullptr, the Session destructor's Delete() call returns early without releasing database_ref_. This could cause issues if GC runs during environment teardown. Release the reference explicitly during DeleteAllSessions() to ensure cleanup happens while the N-API environment is still valid. Build & Release #244: Commit fb283df pushed by mceachen
21m 2s main
main
21m 2s
Build - chore(fmt) Build & Release #243: Commit d6f42ab pushed by mceachen
14m 42s main
main
14m 42s
Build - fix(Session): prevent use-after-free when database is GC'd before session The Session class was using a raw DatabaseSync* pointer without holding a reference to keep the database object alive. When JavaScript lost its reference to the DatabaseSync and GC ran, the database destructor would be called while Sessions still referenced it, causing SIGSEGV on Alpine Linux. Fix: Add Napi::ObjectReference database_ref_ to Session class, matching the pattern used by StatementSync to prevent the same issu... Build & Release #242: Commit a151cb6 pushed by mceachen
3m 0s main
main
3m 0s
Build - chore(fmt) Build & Release #241: Commit b5835fa pushed by mceachen
19m 18s main
main
19m 18s
Build - fix: ensure shell option is set for npm commands on Windows Build & Release #240: Commit 80fa40d pushed by mceachen
23m 38s main
main
23m 38s
Build - chore: remove outdated Node.js SQLite test compatibility documentation Build & Release #239: Commit 5e71100 pushed by mceachen
37m 28s main
main
37m 28s
Build - chore(deps-dev): bump the development group across 1 directory with 4 updates Build & Release #238: Pull request #92 opened by dependabot bot
30m 39s dependabot/npm_and_yarn/development-81d756a982
dependabot/npm_and_yarn/development-81d756a982
30m 39s
Build - chore: update tar package to version 7.5.6 Build & Release #237: Commit 7173728 pushed by mceachen
37m 0s main
main
37m 0s
Build - chore: update better-sqlite3 and typedoc to latest versions Build & Release #236: Commit 5ca322d pushed by mceachen
33m 26s main
main
33m 26s
Build - chore(deps): bump the actions group with 4 updates Build & Release #235: Pull request #91 opened by dependabot bot
26m 39s dependabot/github_actions/actions-d53dc186c3
dependabot/github_actions/actions-d53dc186c3
26m 39s