What happened with AppVeyor greenness?

https://ci.appveyor.com/project/kinke/ldc/history

The commit where red started seems totally unrelated to Win code.

On Saturday, 13 August 2016 at 17:31:03 UTC, Johan Engelen wrote:
> What happened with AppVeyor greenness?
>
> https://ci.appveyor.com/project/kinke/ldc/history
>
> The commit where red started seems totally unrelated to Win code.

We experienced something similar for Mir. They changed their build environment a while ago, might that be your issue?

https://www.appveyor.com/blog/2016/07/16/migration-to-rackspace

AFAIK Ilya wrote the support team, but without response yet :/

On Saturday, 13 August 2016 at 21:14:24 UTC, Seb wrote:
> On Saturday, 13 August 2016 at 17:31:03 UTC, Johan Engelen wrote:
>> What happened with AppVeyor greenness?
>>
>> https://ci.appveyor.com/project/kinke/ldc/history
>>
>> The commit where red started seems totally unrelated to Win code.
>
> We experienced something similar for Mir. They changed their build environment a while ago, might that be your issue?
>
> https://www.appveyor.com/blog/2016/07/16/migration-to-rackspace
>
> AFAIK Ilya wrote the support team, but without response yet :/

https://github.com/appveyor/ci/issues/974

On Saturday, 13 August 2016 at 17:31:03 UTC, Johan Engelen wrote:
> What happened with AppVeyor greenness?
>
> https://ci.appveyor.com/project/kinke/ldc/history
>
> The commit where red started seems totally unrelated to Win code.

Oh, heh, that would be me.  Core.thread is throwing an exception now, that makes no sense.  Feel free to revert my commit, see if that makes a difference.

On Saturday, 13 August 2016 at 22:01:27 UTC, Joakim wrote:
> On Saturday, 13 August 2016 at 17:31:03 UTC, Johan Engelen wrote:
>> What happened with AppVeyor greenness?
>>
>> https://ci.appveyor.com/project/kinke/ldc/history
>>
>> The commit where red started seems totally unrelated to Win code.
>
> Oh, heh, that would be me.  Core.thread is throwing an exception now, that makes no sense.  Feel free to revert my commit, see if that makes a difference.

This surely doesn't make a difference. The AppVeyor guys probably changed something in the last 5 days. I can't reproduce it locally, and I've also never seen that 'OTHER_FAULT'/'Exception: Other' failure reason so far. It could really have something to do with CPUID. In contrast to the Mir guys, we haven't had a problem with the new build environment so far - on the contrary, our AppVeyor times have halved.

> On Saturday, 13 August 2016 at 22:01:27 UTC, Joakim wrote:
>> On Saturday, 13 August 2016 at 17:31:03 UTC, Johan Engelen wrote:
>>> What happened with AppVeyor greenness?
>>>
>>> https://ci.appveyor.com/project/kinke/ldc/history
>>>
>>> The commit where red started seems totally unrelated to Win code.
>>
>> Oh, heh, that would be me.  Core.thread is throwing an exception now, that makes no sense.  Feel free to revert my commit, see if that makes a difference.

dmd-testsuite shows some more failures/crashes for Win64: runnable/eh (RC 9), runnable/test4 (RC 5), runnable/testdstress (RC 9), for both debug and release.

The x86 job is still green. Changing the D compiler to DMD (as used by the x86 job) doesn't help. I still can't reproduce it on my box - up-to-date Win10, VS 2015 Update 3, latest LLVM 3.9.  Running out of ideas...