b9971

ggml-org/llama.cppb9971Jul 12, 2026by github-actions[bot]same day as b9970

What changed since b9970

low risk

Server stream refactor removes a non-thread-safe stop call and reorganizes spipe handling; no model or ggml changes.

Improvements

  • Removed a non-thread-safe rd.stop() call from server_stream, improving thread safety of streaming responses.
  • Relocated server_res_spipe out of server_http_res for cleaner component separation.
  • Reworked server_stream_create_spipe for a clearer, more maintainable implementation.
  • Updated developer documentation to reflect the revised server stream API.

Upgrade notes

This is an internal server refactor with no config or API changes called out in the notes; a drop-in rebuild should suffice. If you maintain out-of-tree patches touching server_stream, server_res_spipe, or server_http_res, expect to rebase against the moved/renamed internals. No migration steps are required for typical users.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

Refactors the server stream functionality to improve thread safety and code organization by removing a non-thread-safe stop call and relocating server components.

Key Highlights

  • Refactored `server_stream` to remove non-thread-safe `rd.stop()` call
  • Moved `server_res_spipe` and improved `server_stream_create_spipe` implementation
  • Updated dev docs for the improved API

Release Notes (excerpt)

<details open> server: refactor server_stream (#25541) * server: refactoring, remove spipe from server_http_res * wip * remove non-thread-safe rd.stop() call * move server_res_spipe * nits * improve server_stream_create_spipe * server-stream: update dev docs for the improved API --------- Co-authored-by: Pascal <admin@serveurperso.com> </details> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories