| Filename | Latest commit message | Latest commit date |
|---|---|---|
|
All checks were successful
Background release validation / Prepare Version (push) Successful in 3s
Background release validation / Fetch latest upstream zapret2 release (push) Successful in 4s
Magisk source guards / guards (push) Successful in 5s
Background release validation / Shell integration tests (push) Successful in 1m18s
Background release validation / Build Android APK (push) Successful in 5m14s
Background release validation / Package Magisk Module (push) Successful in 10s
Owner request 2026-11-16: one heavy build at a time across CT 210 and the Android CI jobs of the Forgejo runner in CT 206 (same US2-7 disks). Where the machine provides us2-build-lock (CT 210), build.sh and the release builder now also take the server-wide lock after this machine's lock: they wait (showing who holds it), refuse on timeout and only warn when the runner cannot be reached. Elsewhere (CI, other machines) nothing changes. The helper process closes fd 9 so it never keeps this machine's lock after the build has exited. tests/release/build-lock.sh stubs the client and checks every answer (HELD, WAIT, TIMEOUT, UNAVAILABLE, no answer, a helper that lingers, a refused build in a live shell); 6/6 mutations caught. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
| .. | ||
| build-lock.sh | ||
| release-channel.sh | ||
| source-policy.sh | ||