Files
nxdns/.gitea/workflows/ci.yml
T
mokhtar c428bc2398
Gates / frontend (push) Successful in 1m18s
Gates / test (push) Successful in 2m27s
Gates / test-aarch64 (push) Successful in 5m39s
Gates / package (push) Successful in 3m34s
Gates / container (push) Successful in 13s
CI / gates (push) Successful in 11m57s
Release / guard (push) Successful in 1m30s
Gates / frontend (push) Successful in 1m6s
Gates / test (push) Successful in 1m28s
Gates / test-aarch64 (push) Successful in 5m41s
Gates / package (push) Successful in 28s
Gates / container (push) Successful in 16s
Release / gates (push) Successful in 7m55s
Release / publish (push) Successful in 10m37s
ci: allow dispatching the gate set, a replaced runner pod kills the job it holds
2026-08-14 02:34:06 +02:00

22 lines
830 B
YAML

name: CI
# CI is nothing but the gate set. Every blocking check lives in gates.yml so
# that release.yml runs the identical set before it publishes anything
# (milestone-14 ruling 7). Nothing may be added here: a check that exists in
# CI but not in gates.yml is a check a release skips.
on:
push:
branches: [master]
pull_request:
branches: [master]
# A runner pod replaced mid-job kills the job it is running: the step's log
# stops dead, no epilogue is written, and Gitea marks the orphan failed.
# Nothing in the tree is wrong when that happens and nothing in the tree can
# fix it, so the gate set must be re-runnable without an empty commit. This
# adds a trigger, not a check — the job list below stays the gate set alone.
workflow_dispatch:
jobs:
gates:
uses: ./.gitea/workflows/gates.yml