Skip to content

RELEASE.md

Latest commit

 

History

History
95 lines (68 loc) · 2.93 KB

RELEASE.md

File metadata and controls

95 lines (68 loc) · 2.93 KB

Releasing mesh-llm

Prerequisites

  • just installed (brew install just)
  • gh CLI authenticated (gh auth status)
  • llama.cpp fork cloned (just build does this automatically)

Steps

1. Build everything fresh

just build

This clones/updates the llama.cpp fork if needed, builds with -DGGML_METAL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF -DLLAMA_OPENSSL=OFF, and builds the Rust mesh-llm binary.

2. Verify no homebrew dependencies

otool -L llama.cpp/build/bin/llama-server | grep -v /System | grep -v /usr/lib
otool -L llama.cpp/build/bin/rpc-server | grep -v /System | grep -v /usr/lib
otool -L mesh-llm/target/release/mesh-llm | grep -v /System | grep -v /usr/lib

Each should only show the binary name — no /opt/homebrew/ paths.

3. Create the bundle

just bundle

Creates /tmp/mesh-bundle.tar.gz containing mesh-llm, rpc-server, llama-server.

4. Smoke test the bundle

mkdir /tmp/test-bundle && tar xzf /tmp/mesh-bundle.tar.gz -C /tmp/test-bundle --strip-components=1
/tmp/test-bundle/mesh-llm --model Qwen2.5-3B --console
# Should download model, start solo, API on :9337, console on :3131
# Hit http://localhost:9337/v1/chat/completions to verify inference works
# Ctrl+C to stop
rm -rf /tmp/test-bundle

5. Commit, tag, push

git add -A && git commit -m "v0.X.0: <summary>"
git tag v0.X.0
git push origin main --tags

6. Create GitHub release

VERSION=v0.X.0
cp /tmp/mesh-bundle.tar.gz /tmp/mesh-llm-${VERSION}-aarch64-apple-darwin.tar.gz
cp /tmp/mesh-bundle.tar.gz /tmp/mesh-llm-aarch64-apple-darwin.tar.gz

gh release create ${VERSION} \
  /tmp/mesh-llm-${VERSION}-aarch64-apple-darwin.tar.gz \
  /tmp/mesh-llm-aarch64-apple-darwin.tar.gz \
  --title "mesh-llm ${VERSION}" \
  --notes "## What's new

- <changelog here>

### Install (macOS Apple Silicon)

\`\`\`bash
curl -fsSL https://github.com/michaelneale/decentralized-inference/releases/latest/download/mesh-llm-aarch64-apple-darwin.tar.gz | tar xz && sudo mv mesh-bundle/* /usr/local/bin/
\`\`\`
"

Two assets are uploaded: one with the version in the name (for pinning), one without (for the latest URL used in the README).

7. Verify the install one-liner works

curl -fsSL https://github.com/michaelneale/decentralized-inference/releases/latest/download/mesh-llm-aarch64-apple-darwin.tar.gz | tar xz && sudo mv mesh-bundle/* /usr/local/bin/
mesh-llm --model Qwen2.5-3B --console

Notes

  • The unversioned asset name (mesh-llm-aarch64-apple-darwin.tar.gz) is what the README's install one-liner uses via the /latest/download/ URL. It must be uploaded with every release.
  • codesign and xattr may be needed on the receiving machine if macOS Gatekeeper blocks unsigned binaries: 
    codesign -s - /usr/local/bin/mesh-llm /usr/local/bin/rpc-server /usr/local/bin/llama-server
xattr -cr /usr/local/bin/mesh-llm /usr/local/bin/rpc-server /usr/local/bin/llama-server