KhronosGroup
Redundant job execution in PRs against main - can we simplify?
@charles-lunarg I note in .github/workflows/ci.yml, it says
on:
push:
pull_request:
branches:
-main
ISTM this is just causing CI to be run twice on every PR against main, which is redundant and slows down the spec update process noticeably since the Windows jobs are very slow to launch. Is there a good reason not to just remove the trigger on PR and leave only the trigger on pushes?
Is there a good reason not to just remove the trigger on PR and leave only the trigger on pushes?
It wouldn't run CI on PRs, which are the only way non-project members can propose changes.
If a PR is coming from a fork rather than a branch, then only the 'pull_request' trigger activates. This workflow is what LunarG typically uses, and also allows for users to push branches up in their own fork with CI active without creating PR's to main.
Also, because forks pull CI runners from a separate pool from the KhronosGroup, CI jobs in forks often start much sooner thank KhronosGroup.
Unfortunately, I do not have a good solution for slow CI startup times.
