Re-enable arm32 windows testing #642
Conversation
|
I would prefer for this to include PR, if we have capacity issues, we can remove the pr bit. |
|
/cc @BruceForstall @echesakovMSFT @dotnet/jit-contrib |
| @@ -229,6 +229,7 @@ jobs: | |||
| helixQueuesTemplate: ${{ parameters.helixQueuesTemplate }} | |||
| osGroup: Windows_NT | |||
| archType: arm | |||
| platform: Windows_NT_arm | |||
There was a problem hiding this comment.
This led us down the path of not actually adding a windows_nt_arm queue. Which is why our ci currently skips windows arm32
There was a problem hiding this comment.
Oops. Maybe we should error out if Queues are empty? Rather than just skipping that step?
That's what happens for Libraries and it catches errors like that.
|
Started https://dev.azure.com/dnceng/public/_build/results?buildId=449917 as there may be regressions. Will disable tests if that is true. |
|
Will this be merged soon because the CoreCLR outerloop build turned out green, or is there more to do? |
|
The coreclr failures look to be the issue fixed by #658 The runtime (Libraries Test Build Windows_NT arm64 Release) failure looks to be an AzDO infra issue. As a result, I think we can merge this in. Anyone feel otherwise? |
|
Well, I'm afraid Jarret's change suffered from the race condition I introduced before the weekend. I'll retry the failed jobs and I'll merge the change in if they look reasonable. Sorry for the delay. |
|
OK, if you say so, let's try that. |
ghost
locked as resolved and limited conversation to collaborators
This also fixes a subtle bug that erroneously lead us to lose coverage on windows arm in ci for a few weeks.