ci: Add continuous_integration dependency#3785
ci: Add continuous_integration dependency#3785arcady-lunarg merged 1 commit intoKhronosGroup:mainfrom
Conversation
|
I don't think this will add very much value: CI jobs mostly succeed, and this just adds latency to every single run. I think it would be better to prune the ci matrix instead, the finding about "gcc" on macos will certainly help us a lot with that. |
https://github.com/KhronosGroup/glslang/actions/workflows/continuous_integration.yml?query=is%3Afailure there have been 137 failure and 531 successes (so that is a 20% failure rate) is the latency something you find that would honestly be a bottleneck, |
|
@spencer-lunarg Could you rebase? I'll approve and merge it if you do. |
69ba099 to
8c134f6
Compare
|
done |
| matrix: | ||
| os: [macos-latest] | ||
| compiler: [{cc: clang, cxx: clang++}] | ||
| cmake_build_type: [Debug, Release] |
There was a problem hiding this comment.
We probably want to keep a macos debug build just in case.
There was a problem hiding this comment.
why? Have we ever seen it fail ONLY on Mac Debug build?
There was a problem hiding this comment.
We came to same conclusion on VVL and other repo that why do we need to worry about the Mac debug build, we want to make sure it works on release to ship, but the number of mac developers on Glslang (or just people on glslang) I can't see why to justify this IMO
There was a problem hiding this comment.
I think the combination of arm64 and mac is useful enough that I would like to run the tests with debug assertions, especially since the tests don't take that long to run and the rate of PRs is a bit lower compared to VVL.
There was a problem hiding this comment.
because glslang is currently not on the "top 5 offenders" of CI action minutes yet, sure will revert
8c134f6 to
57669ea
Compare
In effort to reduce the number of limit KhronosGroup minutes we share across repos, this adds dependency to the Continuous Integration CI to only expensive jobs once quicker/smaller jobs run first