This release includes 1 breaking change for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Summary
AI summaryRemoved the MKL feature causing CI/CD build failures on Linux and Windows.
Full changelog
This hotfix addresses the CI/CD build failure introduced in v0.5.3/v0.5.4 on Linux and Windows platforms.
🔧 Fixes
- Removed the
mkl(Intel Math Kernel Library) feature for Linux and Windows builds. Theintel-mkl-srccrate currently lacks thehgemm_(half-precision general matrix multiply) symbol required by newer versions ofcandle-corefor fp16 tensor operations, which caused a linker failure in the CI environment. - Retained the
x86-64-v3compiler flag (AVX2/FMA instructions), ensuring CPU-bound embedding generation still runs multiple times faster than standard baseline builds. - Retained the Apple
acceleratefeature for macOS (as it correctly supports all required BLAS symbols). - Retained the Rayon/OpenMP thread-pinning fix to prevent tokio worker starvation on high-core-count machines.
Breaking Changes
- Removed `mkl` feature for Linux and Windows builds due to missing `hgemm_` symbol causing linker failure.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About pomazanbohdan/memory-mcp-1file
A self-contained Memory server with single-binary architecture (embedded DB & models, no dependencies). Provides persistent semantic and graph-based memory for AI agents.
Related context
Beta — feedback welcome: [email protected]