Log file Viewer

[00:03:59] <di_codes> Anyone coming for the office hours: I've had a family thing come up and will be pushing it back by 1 hour. Sorry for any inconvenience!

[00:58:06] <di_codes> Welcome to the Warehouse/PyPI livechat, the last of four during this beta!

[00:59:13] <di_codes> Warehouse, the codebase behind the new PyPI, is now in beta <http://pypi.org|pypi.org>, and I'm one of the people working on it. I’m Dustin (@di), a PyPI maintainer/administrator and PyPA enthusiast.

[00:59:31] <di_codes> We intend to make <http://pypi.python.org|pypi.python.org> redirect to Warehouse on Monday, April 16th, and then fully shut down the legacy PyPI site <http://pypi.python.org|pypi.python.org> on Monday, April 30th.

[00:59:48] <di_codes> Our developer roadmap: <http://wiki.python.org/psf/WarehouseRoadmap|wiki.python.org/psf/WarehouseRoadmap>

[01:00:15] <di_codes> We want to hear from you about what you like &amp; what doesn't work for you in the new interface. <http://pyfound.blogspot.com/2018/03/warehouse-all-new-pypi-is-now-in-beta.html#workflows|pyfound.blogspot.com/2018/03/warehouse-all-new-pypi-is-now-in-beta.html#workflows> has a list of stuff we especially want people to test.

[01:00:46] <di_codes> And, as a reminder, participating in this livechat means you agree to abide by the PyPA code of conduct: <http://pypa.io/en/latest/code-of-conduct|pypa.io/en/latest/code-of-conduct>

[01:35:56] <njs> di_codes: I don't have any questions, but I'm excited :-)

[01:37:22] <di_codes> njs: me too :)

[03:58:22] <techalchemy> Hm I am experiencing a weird issue with warehouse

[03:58:41] <techalchemy> I was just checking the json api as I do from time to time

[03:59:29] <techalchemy> and I found that I can hit https://pypi.org/pypi/funcsigs/1.0.2/json which redirects to => https://pypi.org/project/funcsigs/1.0.2/json

[04:00:07] <techalchemy> but I can't hit either https://pypi.org/pypi/funcsigs/json or https://pypi.org/project/funcsigs/json

[04:00:30] <techalchemy> just want to confirm that's an actual bug and not a documentation issue

[04:15:13] <di_codes> techalchemy: <https://pypi.org/pypi/funcsigs/1.0.2/json> does not redirect for me, and <https://pypi.org/pypi/funcsigs/json> correctly resolves. <https://pypi.org/project/funcsigs/json> does not exist and won’t redirect

[04:16:34] <di_codes> what are you using to hit these endpoints?

[04:21:14] <techalchemy> web browser

[04:21:51] <techalchemy> di_codes: they work now, but they were not working before

[04:22:20] <techalchemy> if i knew they were transient I would have done more debugging, I'll catch more info if I see it again

[04:22:21] <techalchemy> thanks

[04:22:37] <di_codes> techalchemy: <https://pypi.org/pypi/funcsigs/1.0.2/> will redirect to <https://pypi.org/project/funcsigs/1.0.2/>, maybe there was some confusion?

[04:22:39] <di_codes> np

[04:23:15] <techalchemy> nah I double checked before i linked (probably)

[04:23:20] <techalchemy> :D

[04:23:29] <techalchemy> I'm 100% sure that I think I verified

[14:19:00] <ronny> jaraco: ping? i like to drop support for setuptools versions that dont support modern version parsing in setuptools_scm, but im not sure how to go about the details of that move

[14:26:31] <jaraco> @ronny: Given the way setuptools_scm is typically resolved on-demand at build-time by easy_install (and not pip), and since setuptools is already loaded, I think the best you can do is fail if the required setuptools isn’t present. Given that support has been warned against for a long time, I’d simply remove the compatibility support, cut a new backward-incompatible release, and let the users pin to older setuptools_scm

[14:26:31] <jaraco> if they have old setuptools.

[14:26:53] <jaraco> (package maintainers would have to pin, I guess).

[14:27:17] <jaraco> But my suspicion is that most environments are updating to at least setuptools 28.8 these days.

[14:42:23] <ronny> jaraco: ok, thaqnks, then i'll prepare arelease - including support for semver based increments and removal of support for old setuptools

[14:42:38] <jaraco> !m ronny

[14:42:38] <pmxbot> you're doing good work, ronny!

[15:42:40] <techalchemy> lol ok so di_codes I figured it out, if I hit https://pypi.org/pypi/funcsigs/json it works fine

[15:42:53] <techalchemy> but if I hit https://pypi.org/pypi/funcsigs/json/ it returns a redirect to /project/funcsigs/json/

[15:48:59] <mgedmin> lol

[15:58:50] <di_codes> techalchemy: ah yeah, that’s bugged me before too. i’d merge a PR that fixes that behavior

[16:00:14] <techalchemy> I'd make a PR that fixes it if i had time :D

[16:18:38] <ronny> jaraco: ping, may i rope you into giving a review to https://github.com/pypa/setuptools_scm/pull/240 - it adds support for branch metadata and adds a version scheme based on simplified semver (incrementing patch on normal branches and minor on feature branches)

[16:32:23] <jaraco> @ronny I’ll try to look today.

[16:35:48] <ronny> jaraco: thanks

[16:36:29] <ronny> jaraco: im primaryly looking for strucutral scrinity and edges i missed for making change easy

[16:41:40] <ronny> jaraco: ah right, whats the best way for setuptools_scm to require setuptools with a certain featureset, is the install_requires sufficient or do i need to put a require or a feature check into the code?

[16:44:30] <jaraco> @ronny: I’m not sure. Ideally install_requires would be sufficient, and it’s quite possible that if installed by easy_install/setuptools (as part of a package’s setup_requires), setuptools would reject if the version required is newer than what’s being used. What I’d be afraid of is if setuptools tried to upgrade itself in place as a result of that requirement.

[16:45:08] <jaraco> My best recommendation would be to omit it (leave the requirement implicit), test a few scenarios to verify the behavior, or write imperative checks into the setuptools_scm code.

[16:49:18] <ronny> jaraco: i see, in that case i intend to fail if the version class is not availiable (just elevating the warning

[16:50:16] <ronny> i'll use the warnings system and add a acceptance test, so integrators can opt out of it via pythonwarnings

[16:50:42] <jaraco> Sounds like a plan.

[16:53:03] <ronny> thanks for the input

[17:02:23] <ronny> oO

[17:02:31] <ronny> pip can no longer install really old setuptools

[20:20:25] <thea> heads up I'm trying to get syntax highlighting for gfm in today.

[21:25:30] <thea> whenever (any PyPA member) has a free moment: https://github.com/pypa/readme_renderer/pull/82

[22:04:03] <di_codes> thea: will take a look at that this weekend!

[22:05:42] <thea> Sweet. I'll cut a release after. :)