Tests are failing on appveyor

[pfmoore]

The test suite is currently failing on Windows. The appveyor builds are showing as red. I have a patch which I will submit once the CI has run on my clone. Note that the patch will still fail on the main repo, this is because of #302 but when #304 is applied, the appveyor tests all pass.

@bitprophet if you get CI test failures on Windows and need help with fixes, please feel free to ping me. If the only thing stopping us having a clean Windows build is lack of resource with Windows knowledge, I'm happy to help out.

[pfmoore]

OK, #306 created. In conjunction with #304, tests should start passing on Windows again.

[bitprophet]

Previously, tests on appveyor would fail after ~9-12 minutes; the build including #306 (https://ci.appveyor.com/project/bitprophet/invoke/build/1.0.208/job/0u2iid1i4um7w60g) is now up to 20+ minutes...but...it seems to be stuck on a single test, whichever one it's running first under Program.core_args (which is especially vexing as that stuff should all be basic, non platform specific unit tests).

Hoping it's just really crummy output buffering or something; guess we'll see, depending on whether appveyor has a timeout.

EDIT: almost 40 minutes now. Even if this succeeds it doesn't bode well for utility of Appveyor, heh. Tempted to cancel on assumption of getting stuck, but really curious now as to what point Appveyor will give up.

Also double checked and the test it's presumably running right now is indeed a very basic, no platform stuff involved at all, unit test :( I see nothing in there that would hang like this.

EDIT 2: Looks like it's a 60 minute timeout most likely, per this doc:

All plans have 60 minutes quota per build job.

[bitprophet]

Yup it hit that timeout. Rest of jobs in that build are still queued so we'll see if it's a fluke or what. Gonna put aside and go back to other things for now.

[pfmoore]

I wonder if appveyor is running the jobs without a console somehow, and that's affecting the msvcrt stuff. I'll try to see if I can work out what's going on.

[pfmoore]

OK, worked it out - see #316

[bitprophet]

Sweet, thanks for poking at that. Just merged, we'll see if my recent merges broke anything else I guess. In a while. Poor slow appveyor =/ I will check back here before I cut 0.12.1. (Soon, now that I finally finished #308, sigh).

[bitprophet]

Canceled the intermediate jobs since they also wouldn't pass; the job after I merged #316 seems to have passed, at least on the first interpreter matrix cell: https://ci.appveyor.com/project/bitprophet/invoke/build/1.0.213/job/8k308muc61bed5jt

We don't/can't run the integration tests on here so I can't tell if my "nested invoke" bugfix works on Windows, but as long as the more normal test cases work fine, that's plenty good enough for me...

[bitprophet]

Yup all passed.