-
Notifications
You must be signed in to change notification settings - Fork 3.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Spread the Windows test load between Azure & Appveyor #6790
Spread the Windows test load between Azure & Appveyor #6790
Conversation
Appveyor is angry at you Xavier. |
This looks suspiciously like pypa/virtualenv#1339 You need to be very careful to get a completely up to date virtualenv when using 3.7, as behaviour changes within the (internal details of) minor versions of 3.7 repeatedly broke virtualenv. I think we got all the edge cases dealt with, but if anyone finds a new one, it'll need addressing (although @gaborbernat is rewriting virtualenv to use the core venv mechanisms, so hopefully this whole mess will be behind us soon). |
Appveyor seems to be quite sensitive indeed.
yup, it looks like it. I moved 3.7 build to azure to avoid this (or get the same error on Azure ^^). |
fe7f2be
to
a66891b
Compare
a66891b
to
30b7720
Compare
Ah, optimism. ;) @pfmoore this was the issue with the behavior varying for patch versions of CPython 3.7, right? FWIW, pip "officially" only supports the latest patch versions of each non-EOL minor version. Maybe we should use the newest CPython minor version on CI (via pyenv or equivalent), with caching to optimize performance? |
Well Azure seems more sympathetic and tests seem to go smoothly.
|
@pradyunsg Looks like the issue is resolved. But yes, in theory we could force use of the latest version. Seems like a lot of bother though. Making sure we use the latest virtualenv should be enough. Personally, I'm a little frustrated that this stuff got changed in CPython minor releases without proper deprecation - but it is internal details, so it's hard to get too upset (it's just that there's no supported way to implement virtualenv, using internals is necessary - I'm so looking forward to the rewrite when we can finally put all of this rubbish behind us) . |
Fair enough. 🙃 |
Merging since no one is opposed to doing this. |
Cf #6791 for the rationale