Log file Viewer

[00:35:56] <tchaypo> well, got pep-496 updated, but that ended up taking longer than expected. Will tackle env-markers-in-install_requires writeup this afternoon

[00:49:05] <lifeless> tchaypo: I'm reviewing it now

[00:57:41] <lifeless> stoppnig for now, gym time

[02:43:04] <lifeless> tchaypo: finished it I think

[03:30:53] <Nakato> dstufft: I've got a test in my current patch for constraints, and it fails, which is correct, do you prefer that to be put in now as an "xfail" or should I yank the test out and add it in as I put a fix in?

[04:56:12] <tchaypo> lifeless: i love it when i update things and make them worse

[04:58:54] <lifeless> tchaypo: :/

[05:51:49] <tchaypo> whee

[05:52:12] <tchaypo> https://www.python.org/dev/peps/pep-0345/#environment-markers doesn’t talk about install_requires, it talks about Requires-Dist

[05:52:27] <tchaypo> how does that map to *_require(|s)...

[05:56:27] <tchaypo> oh, https://www.python.org/dev/peps/pep-0426/#mapping-dependencies-to-development-and-distribution-activities makes this pretty clear

[05:56:47] <tchaypo> test_requires is implied to be identical with the extra named :test:

[07:53:18] <lifeless> tchaypo: except, that 426 was aspirational

[07:53:31] <lifeless> tchaypo: and last I heard setuptools doesn't consider test_requires to be == to test

[07:54:13] <lifeless> tchaypo: you'll note that run_requires which isn't install_requires but might be install_requires

[07:54:47] <lifeless> tchaypo: PEP-426 is entirely focused on replacing the METADATA file

[07:55:18] <lifeless> tchaypo: also actually I think you misread it

[07:55:29] <lifeless> tchaypo: :test: says 'process test_requires as well'

[07:55:58] <lifeless> tchaypo: which is not the same thing as equality

[08:04:58] <tchaypo> After walking and thinking, I'm wondering if I've gotten too focused on one particular solution and I'm missing the bigger picture

[08:06:35] <tchaypo> I've been thinking about how to show-horn env markers into tests_require/install-requires/setup_requires as passed in to the setup() function

[08:07:12] <tchaypo> But the original problem was to be able to get env markers in test requirements, which is a bit broader.

[11:05:09] <lifeless> the bigger picture is to make environment markers a) easy to work with and b) available consistently across all the tools. The glaring omission today is in the setup_requires,tests_require parameters to setup() in setuptools.