Log file Viewer

[01:46:17] <tchaypo> lifeless: I’ve put a stub repo that reproduces the problem at https://github.com/jamezpolley/setuptools-380-testpackage

[01:46:52] <tchaypo> lifeless: it behaves differently depending on whether or not pbr is installed into the repo

[01:48:52] <tchaypo> if pbr isn’t installed (but gets loaded as a setup-requires), the env-marked requirements get written into requires.txt, but don’t get installed. If I pip install pbr into the venv, we stumble on the “invalid maker” error

[01:49:38] <tchaypo> I’m wondering what would make pbr behave differently depending on whether it’s pre-installed or loaded as a setup-requires. I’m 90% sure that I confirmed on the plane that either way the same version number is being loaded

[05:48:43] <lifeless> tchaypo: interesting

[06:34:29] <tchaypo> More interesting.

[06:34:57] <tchaypo> a search for the “Invalid environment marker” string only finds it at setuptools/dist.py:106

[06:35:21] <tchaypo> but when I set a breakpoint there (on the file in the venv, and on the system version just in case that’s being invoked), the program ends without hitting the breakpoint

[07:33:19] <tchaypo> … oh. of course. I think this also explains why I’m seeing a sys.exit() having happened rather than a trace - I think something is shelling out and calling setup.py again, and it’s the re-invokation that is failing