Log file Viewer

[12:37:22] <Xorg> If I want to takeover a package thats not written by me, but stopped supporting it

[12:37:25] <Xorg> Any procedure ?

[12:46:22] <zaytsev_> hello

[12:46:31] <Wooble> hi.

[12:47:33] <zaytsev_> ubuntu 14.04 / latest pip, am i correct in my suspicions that pip can't understand the difference when run via python -m pip and python-dbg -m pip and picks same wheels from cache for both :-( ?

[12:48:35] <zaytsev_> at least i see no other explanation for mysterious import errors like "_psutil_linux.so: undefined symbol: Py_InitModule4_64"

[12:54:40] <zaytsev_> now going to retry with separate --cache-dir, but if that's true, it's really evil :-(

[13:11:22] <zaytsev_> yeah, what a fail

[13:11:29] <zaytsev_> i wonder if it's worth reporting a bug

[17:59:46] <ionelmc> zaytsev_: debug mode don't change the wheel tag unfortunately

[18:19:18] <ionelmc> zaytsev_: fyi https://mail.python.org/pipermail/distutils-sig/2015-April/026086.html

[18:57:40] <glyph> did download stats on pypi stop working?

[18:59:05] <glyph> because this curve looks ... unrealistically steep: https://gist.github.com/3c31bb9f2a3219e6990f4c76f4645697

[19:06:09] <dstufft> glyph: maybe Twisted just really went downhill in 2016

[19:06:16] <dstufft> glyph: stats are probably broken again :[

[19:06:26] <dstufft> I hate legacy pypi

[19:06:40] <glyph> dstufft: too bad you folks made The Second System Mistake :(

[19:06:58] <glyph> dstufft: do I need to link you to some jwz and joel on software posts

[19:07:52] <dstufft> glyph: I stand behind my decision to Warehouse. Legacy was too broken to save and working on it was too painful to actually land changes.

[19:12:38] <dstufft> (I only half way agree with joel on that, and I think a lot of his points are nonsense)

[19:14:39] <ionelmc> glyph: it's more like 4th system, warehouse being rewritten couple times, right dstufft?

[19:15:43] <dstufft> But, rest assured, this will be the sixth time we have destroyed---- rewritten it, and we have become exceedingly efficient at it.

[19:16:07] <ionelmc> then again, the only way to prove dstufft wrong is to actually do the work of maintaining pypi, successfully :-)

[19:17:49] <ionelmc> i enjoy rewriting stuff as well but i think glyph has a point here

[19:19:54] <elarson> glyph: one bit of reality is that dstufft isn't paid to maintain pypi, so if the decision is pypi bit rot vs. getting a new system that is maintained, then the latter is probably a better option.

[19:20:44] <glyph> sorry I was not trying to make a point :(

[19:20:56] <ionelmc> elarson: i thought he was ... not anymore?

[19:20:57] <glyph> the question is really just when can the cut-over be made

[19:22:10] <dstufft> elarson: techincally I am paid ;) but not specifically to maintain PyPI, just to "improve packaging" so I can work on PyPI or not as I want (although not being paid vs being paid isn't really that important, I can always get a different job if I don't want to maintain it anymore)

[19:23:29] <elarson> dstufft: yeah, just making the point that when no one is obligated to work on something, then if someone wants to do a rewrite, that is better nothing

[19:24:16] <dstufft> glyph: as far as when- Idk. One of the things we're blocked on is some new UI stuff and Nicole- who is not paid- hasn't had spare cycles to work on it atm.

[19:24:38] <ionelmc> is the search bug still not solved?

[19:24:50] <ionelmc> i think that's worse than any download count issue

[19:24:59] <ionelmc> who even cares about download count!

[19:25:53] <elarson> ionelmc: what?!?! you don't put it on your résumé?

[19:26:45] <ionelmc> elarson: download counts in resume?

[19:26:58] <ionelmc> i suspect a pun

[19:27:03] <ionelmc> :-)

[19:27:22] <ionelmc> i think other things _count_ more in a resume

[19:27:29] <elarson> !m ionelmc

[19:27:30] <pmxbot> you're doing good work, ionelmc!

[19:28:35] <dstufft> Hmm

[19:30:14] <dstufft> I wonder how bad of an idea it would be to move legacy pypi to legay.pypi.python.org and just deploy warehouse missing features. The stuff that is missing affects authors more than other kinds of users and then, only via the webui not via setup.py or twine.

[20:10:19] <Norn> What characters are allowed in a package name?

[20:12:08] <dstufft> Norn: alpha numeric, - _ and .

[20:12:16] <dstufft> must begin and end with an alpha numeric

[20:12:29] <dstufft> case insensitive, and - _ and . are equivelant as well

[20:12:35] <Wooble> (assume you mean "on pypi"

[20:13:42] <Wooble> (in theory a Python package can use any charactersm but it's going to be obnoxious to import it if it's not a valid python identifier :) )

[20:15:48] <Norn> Yes, sorry - meant on PyPI. Couldn't find a definite source of truth - thank you! :)

[20:16:09] <dstufft> If you want a source of truth, I think it's uhhh PEP 501?

[20:16:23] <dstufft> no that's not it

[20:16:39] <dstufft> 503

[20:17:24] <Norn> There we go, thanks again!

[20:18:24] <Norn> "The name should be lowercased with all runs of the characters . , - , or _ replaced with a single - character." Does this mean it can't have dots or underscores?

[20:18:42] <Norn> Or is there a difference between the concept of a normalized name and a regular name?

[20:19:19] <dstufft> Norn: oh, right 503 is just the normalization

[20:19:44] <dstufft> 503 documents the cannonical form of the name, so that Django-tWo normalizes to django-two

[20:25:19] <Norn> I see

[20:28:46] <zaytsev_> ionelmc: thank you very much, but the conclusion is: it's known and isn't going to be fixed, not worth reporting, do i understand this correctly?

[20:29:49] <ionelmc> zaytsev_: it can be fixed, someone just needs to spend time to do it

[20:31:41] <zaytsev_> ionelmc: sure, just like 200 other bugs in pip ^__^ too bad i can't afford to invest time in fixing it, at least not at the moment, sadly. so, not worth filing an issue?

[20:33:03] <ionelmc> i guess not, it's not a simply pip issue anyway