Publish packages as a Forgejo release on v* tags #42

Merged
littlefox merged 2 commits from issue-41-releases into main 2026-10-17 12:12:15 +00:00
Owner

Closes #41

On a v* tag, packaging.yml builds and checks the four distros as before. A new release job then creates the Forgejo release with:

  • the Arch .pkg.tar.zst, the Debian .deb, and the Fedora .rpm and .src.rpm;
  • SHA256SUMS;
  • install notes (packaging/release-notes.md), with NixOS through the tagged flake.

The upload is done by tools/publish-release.py (stdlib only, safe to re-run).

After merging, I'll tag v0.1.0 on main to publish the first release.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8

Closes #41 On a `v*` tag, packaging.yml builds and checks the four distros as before. A new `release` job then creates the Forgejo release with: - the Arch .pkg.tar.zst, the Debian .deb, and the Fedora .rpm and .src.rpm; - SHA256SUMS; - install notes (packaging/release-notes.md), with NixOS through the tagged flake. The upload is done by tools/publish-release.py (stdlib only, safe to re-run). After merging, I'll tag v0.1.0 on main to publish the first release. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8
Publish packages as a Forgejo release on v* tags (D-027)
Some checks failed
ci / desktop (push) Has started running
ci / e2e (push) Failing after 9m26s
ci / check (push) Successful in 10m5s
packaging / arch (push) Successful in 12m38s
packaging / nix (push) Successful in 14m49s
packaging / fedora (push) Successful in 18m27s
packaging / debian (push) Successful in 29m32s
packaging / release (push) Has been skipped
db8780afd5
There were no releases: the packages only existed as workflow artifacts,
which expire. Now a version tag runs the packaging workflow, and a final
job creates the release with the Arch, Debian and Fedora packages,
SHA256SUMS, and install notes (NixOS through the tagged flake).

Refs #41

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8
kom-relay: finish the WebSocket close handshake before dropping it
All checks were successful
ci / e2e (push) Successful in 8m33s
ci / check (push) Successful in 9m10s
ci / desktop (push) Successful in 12m3s
2ee4def1fa
After closing with 4001 (no TLS) the relay dropped the socket at once.
On a busy CI runner the browser then saw an abnormal close (1006)
instead of 4001, showed a generic error instead of the plain-text
question, and tls.spec.ts timed out. Now the relay waits (up to 5 s) for
the browser's close reply.

Refs #39

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XJW2ocwLunyaLiUwCBHEP8
Sign in to join this conversation.
No description provided.