Consolidate OpenAI server: unify implementations, add multi-protocol support by bernardladenthin · Pull Request #243 · bernardladenthin/java-llama.cpp

Consolidate OpenAI server: keep JDK/streaming OpenAiCompatServer, drop NanoHTTPD

claude

committed

feat(server): add /infill + tool/usage correctness for IDE agent backends

claude

committed

feat(server): rerank + structured outputs + vision flag (M batch) and docs

claude

committed

feat(server): Ollama-native API surface (/api/version,tags,show,chat)

claude

committed

feat(server): Anthropic /v1/messages + OpenAI /v1/responses protocol shims

claude

committed

feat(server): add GET /props + Ollama POST /api/generate

claude

committed

docs(TODO): sync OpenAI-server status — full route list + remaining gaps

claude

committed

test(server): gated integration round-trips for the alternative protocols

claude

committed

test(server): extend protocol-translator coverage; fix Anthropic tool_result-only turn

claude

committed

test(server): gated integration round-trips for embeddings, rerank, completion/infill/generate

claude

committed

fix(server): clear spotbugs fb-contrib/findsecbugs findings on the consolidated server

claude

committed

test(server): use MEAN pooling for the OAI embeddings integration test

claude

committed

fix(server): close streaming responses in a Sonar-recognized way; drop opaque-field toString

claude

committed

fix(server): synchronize streamed writes on an owned final lock (SonarCloud S2445)

claude

committed

build(server): compile upstream server-http.cpp + cpp-httplib into libjllama

claude

committed

build(webui): build llama.cpp WebUI once in CI, share to all native builds

claude

committed

fix(server): clear SonarCloud gate — close resources + harden CI curl/npm

claude

committed

feat(server): propagate parallel_tool_calls across all OpenAI-compatible surfaces

claude

committed

build(ci): cap macOS build parallelism + wire shared sccache/Depot compiler cache

claude

committed

docs: credit Depot build cache (README badge) + document the cache in CLAUDE.md

claude

committed

build(ci): phase 2 — sccache/Depot cache for the dockcross + Linux-host builds

claude

committed

fix(ci): enforce HTTPS on the sccache download curl (SonarCloud Security gate)

claude

committed

fix(ci): disable sccache on the dockcross/Linux-host builds (sccache crashed in-container)

claude

committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Consolidate OpenAI server: unify implementations, add multi-protocol support#243

Consolidate OpenAI server: unify implementations, add multi-protocol support#243
bernardladenthin merged 23 commits into
mainfrom
claude/clever-clarke-6a6vvg

Commits on Jun 20, 2026

Consolidate OpenAI server: keep JDK/streaming OpenAiCompatServer, drop NanoHTTPD

feat(server): add /infill + tool/usage correctness for IDE agent backends

feat(server): rerank + structured outputs + vision flag (M batch) and docs

feat(server): Ollama-native API surface (/api/version,tags,show,chat)

feat(server): Anthropic /v1/messages + OpenAI /v1/responses protocol shims

feat(server): add GET /props + Ollama POST /api/generate

docs(TODO): sync OpenAI-server status — full route list + remaining gaps

test(server): gated integration round-trips for the alternative protocols

test(server): extend protocol-translator coverage; fix Anthropic tool_result-only turn

test(server): gated integration round-trips for embeddings, rerank, completion/infill/generate

fix(server): clear spotbugs fb-contrib/findsecbugs findings on the consolidated server

test(server): use MEAN pooling for the OAI embeddings integration test

fix(server): close streaming responses in a Sonar-recognized way; drop opaque-field toString

fix(server): synchronize streamed writes on an owned final lock (SonarCloud S2445)

build(server): compile upstream server-http.cpp + cpp-httplib into libjllama

build(webui): build llama.cpp WebUI once in CI, share to all native builds

fix(server): clear SonarCloud gate — close resources + harden CI curl/npm

feat(server): propagate parallel_tool_calls across all OpenAI-compatible surfaces

build(ci): cap macOS build parallelism + wire shared sccache/Depot compiler cache

docs: credit Depot build cache (README badge) + document the cache in CLAUDE.md

build(ci): phase 2 — sccache/Depot cache for the dockcross + Linux-host builds

fix(ci): enforce HTTPS on the sccache download curl (SonarCloud Security gate)

fix(ci): disable sccache on the dockcross/Linux-host builds (sccache crashed in-container)