Ignore merged test markers for tests without execution script#85441
Ignore merged test markers for tests without execution script#85441trylek wants to merge 1 commit intodotnet:mainfrom
Conversation
|
Tagging subscribers to this area: @hoyosjs Issue DetailsThis delta should fix the remaining issues discussed on the PR thread by filtering out those Thanks Tomas P.S. I'm going to be OOF on vacation Thursday thru Sunday in a recording studio session with my music band. I have validated this PR by putting the commit on top of the above Ivan's PR, please see https://dev.azure.com/dnceng-public/public/_build/results?buildId=254082&view=results I likely won't be able to merge this in before I sign off for the rest of the week. Please feel free to do anything with my change, either by merging it in assuming it works or by adding it as an extra commit on top of Ivan's change.
|
|
Hmmm.... @trylek I'm not sure it's working. If I look at the And look at all the related log files using this Kusto query: It shows HardwareIntrinsics_Arm_r and HardwareIntrinsics_Arm_ro jobs, and the console log for the former shows: |
|
Maybe someone can explain to me: which step in the CI system creates the per-test wrapper scripts (e.g., HardwareIntrinsics_Arm_r.cmd above, for merged tests, and similarly for non-merged tests). I've never been able to figure out where they are created. Who invokes CLRTest.Execute.targets :: GenerateExecutionScript ? [Edit] So it looks like the wrapper scripts are created during the test build process here: runtime/src/tests/Directory.Build.targets Lines 266 to 270 in 318186b but NOT if [Edit 2] By experimentation, it appears to be the |
Agree with the fact that MSBuild output should be clearer. But I'm a bit confused about your overall review Bruce, since you dismissed a "stale" comment. Do you mean the problem is still happening in the CI even with Tomas' changes? |
yes |
Ok I will take a look then. Thanks for finding that. |
|
I've been looking at this. I think a simple change might fix it: #85476 |
|
@BruceForstall, this can be closed in favor of your PR, right? |
yes |
This delta should fix the remaining issues discussed on the PR thread
#85183 (comment)
by filtering out those
MergedTestAssemblymarkers where the entire merged test wrappers got filtered out based onCLRTestTargetUnsupportedor a similar property (a recently hit previously unseen case).Thanks
Tomas
P.S. I'm going to be OOF on vacation Thursday thru Sunday in a recording studio session with my music band. I have validated this PR by putting the commit on top of the above Ivan's PR, please see
https://dev.azure.com/dnceng-public/public/_build/results?buildId=254082&view=results
I likely won't be able to merge this in before I sign off for the rest of the week. Please feel free to do anything with my change, either by merging it in assuming it works or by adding it as an extra commit on top of Ivan's change.