Upgrade llama.cpp from b10319 to b10327

[bernardladenthin]

Summary

• Upgrade llama.cpp pinned version from b10319 to b10327, following on from the b10310→b10319 upgrade merged as Upgrade llama.cpp to b10319: migrate TTS from OuteTTS to Qwen3-TTS #381.
• Two-step walk (b10319→b10322→b10327), each step's diff reviewed against common/, include/, tools/mtmd/mtmd.h, tools/mtmd/mtmd-helper.h, tools/server/*.h, tools/server/*.cpp, CMakeLists.txt, and vendor/sheredom/subprocess.h.
• No project-source changes required — both steps were internal-only on the upstream side:
  • b10319→b10322: internal GPU-backend tweaks (ggml-metal, ggml-sycl) plus a new WebUI @-mention picker feature.
  • b10322→b10327: internal GPU-backend tweak (ggml-cuda) plus a large WebUI contenteditable chat-form rewrite. The one TTS-adjacent change (tools/tts/tts.cpp's reference-tool timing/logging) uses an out-parameter that mtmd_helper::gen_audio::get_output already had before this range (tools/mtmd/mtmd-helper.h has zero diff across the whole range) — tts_engine.cpp's call is unaffected.

Test plan

• Affected unit / integration tests pass locally
  • Full cmake --build with -O3 on the b10327 checkout; ctest 482/482 passing, no undefined references.
  • All 6 local patches re-verified and apply cleanly at both intermediate steps and the final target.
• CI is green on this branch (local equivalent verified; GitHub Actions CI runs on push)
• Docs / CHANGELOG updated where applicable
  • docs/history/llama-cpp-breaking-changes.md gained rows for both steps.

Related issues / PRs

• Follows on from Upgrade llama.cpp to b10319: migrate TTS from OuteTTS to Qwen3-TTS #381 (b10217→b10319, including the Qwen3-TTS migration).

Checklist

• I have read CONTRIBUTING.md and CODE_OF_CONDUCT.md
• My commits follow Conventional Commits
• No security-sensitive changes

https://claude.ai/code/session_01QyzMNtQf878jjtLc6KCxVv

---

Generated by Claude Code

[claude]

Review Summary

This PR successfully upgrades llama.cpp from b10319 to b10327 via a methodical two-step walk (b10319→b10322→b10327), following the well-documented upgrade process in CLAUDE.md.

Version Consistency: All 4 required version-pin locations updated correctly

• CLAUDE.md: Line 9 (project description)
• README.md: Badge and link reference
• llama/CMakeLists.txt: GIT_TAG in FetchContent (line 176)
• LlamaCppVersion.java: Compile-time constant and Javadoc examples

Documentation: docs/history/llama-cpp-breaking-changes.md properly updated with step-by-step analysis for both intermediate and final targets, clearly documenting: changes analyzed, no-change zones, and verification status.

Upstream Compatibility:

• b10319→b10322: GPU backend internals + WebUI features only; no project-source impact
• b10322→b10327: The one TTS-adjacent change (tools/tts/tts.cpp) uses optional out-parameter already existing in mtmd_helper::gen_audio::get_output before this range — tts_engine.cpp's call is backward-compatible and unaffected
• Zero diff confirmed across critical zones: common/, include/, tools/mtmd/mtmd.h, tools/mtmd/mtmd-helper.h, tools/server/.h, tools/server/.cpp, CMakeLists.txt, vendor/sheredom/subprocess.h

Build Verification:

• All 6 local patches re-verified and apply cleanly at both intermediate and final target
• Full build: cmake --build with -O3 on b10327, no undefined references
• C++ tests: 482/482 passing, no regressions

Commit hygiene: Two commits following incremental-review pattern with detailed messages and co-authored credits present.

Ready to merge. ✓

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud