windows CI is flaky

[johnynek]

see: https://travis-ci.org/bazelbuild/rules_scala/jobs/531959514#L436

cc @majcherm-da

this has shown up in: #748 and #745

[johnynek]

I've seen this several times...

I wonder if it is flaky locally or if travis is flaky for windows.

[ittaiz]

I've seen another different issue of windows flakiness attached here (I actually saw two different ones but don't have the log of the other)
failing_windows_travis_log.txt
@majcherm-da do you think you'll have some capacity to take a look?

[johnynek]

I think proto tests should just be disabled on windows until someone interested wants to invest the time to get them passing.

[ittaiz]

+1. We should probably decide that this is the policy for most tests that flake out after some time. The unification Andy sent will probably make this a bit more complicated bash setup wise

…

On Fri, Aug 16, 2019 at 1:05 AM P. Oscar Boykin ***@***.***> wrote: I think proto tests should just be disabled on windows until someone interested wants to invest the time to get them passing. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#749>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKQQFY6TZCPC7WKRI7ETO3QEXHJRANCNFSM4HMTFQUA> .

[andyscott]

I prefer that we avoid hacking the current script--

I intend to split apart the test scripts so it's easy to tag which tests run for each configuration. As part of that split, I'll omit the windows failing tests from the windows CI.

[ittaiz]

👍🏽 One note though- long term vision IMHO should be to move the bash tests inside Bazel using bazel-integration-testing library. If you’re doing a small medium effort that’s ok but if you’re going to be doing a large one then maybe we should reconsider investing in moving over to it.

…

On Sun, 18 Aug 2019 at 3:57 Andy Scott ***@***.***> wrote: I'd prefer that we wait-- I intend to split apart the test scripts so it's easy to tag which tests run for each configuration. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#749>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKQQFYCRLRNSF5IYX5JG2TQFCM5ZANCNFSM4HMTFQUA> .