Windows compatibility

[dsludwig]

Replaces #46

Includes an appveyor configuration for Windows CI: https://ci.appveyor.com/project/dsludwig/conda-pack/

There's an outstanding issue when the archives are extracted on a case-sensitive file-system, but I didn't want to add that complexity for what is likely not a common case. I've included an extensive comment that explains the issue in more detail.

[jcrist]

Before I review (hopefully tomorrow), could you quick fix the merge conflicts that have popped up? This will make it easier to review, and get testing setup on the main branch. Thanks for all your work here.

[dsludwig]

@jcrist I've resolved the latest merge conflicts with master.

Passing build on Windows is here: https://ci.appveyor.com/project/dsludwig/conda-pack/build/1.0.13

[certik]

@dsludwig, the Travis tests fail. Do you know how to fix that?

[dsludwig]

Thanks for the ping @certik, I found the problem in the tests. It's passing now!

[certik]

The changes on linux seem fine to me, and the linux tests pass on Travis, so I would say it's ok. If something broke, we should improve the linux tests.

Regarding the changes on Windows, they look ok as the first iteration, and so I suggest to merge it, so that the appveyor tests get enabled, and so we can then send subsequent PRs and they will get tested on Windows.

@jcrist would you please mind reviewing it and merging?

[certik]

@jcrist ping.

[mcg1969]

AppVeyor is passing here, but holy cow is it slow. pip itself seems to be a major culprit in the slowness, and I'm not sure why.

[mcg1969]

OK, I spoke to soon; @dsludwig, it looks like it's not passing with Python 3

[mingwandroid]

Yes

Ah ok then sorry for the noise. Do what you will with them (within reason!)

[mcg1969]

Per @jcrist's request, #54 has been incorporated.

[mcg1969]

Last test failure was a flake issue. Functional tests passing.

[mcg1969]

Now the only test that doesn't run on Windows is test_keyboard_interrupt. For test_formats, I copied the relevant files/trees instead of symlink to them, so the result is effectively the same as if I did check(..., False).

[jcrist]

@dsludwig and @mcg1969, thanks for all the work here, and for being patient with my slow review time. This is in.

[certik]

Thanks everybody!

[fdabek1]

@jcrist Any idea when this will be released? I am waiting eagerly for the next release on conda-forge so that I can utilize this without installing from source.

[jcrist]

I expect to issue a release sometime in the next week.

[github-actions]

Hi there, thank you for your contribution!

This pull request has been automatically locked because it has not had recent activity after being closed.

Please open a new issue or pull request if needed.

Thanks!