pmxbot IRC Log Viewer

[13:37:37] <ronny> hi

[13:38:45] <ronny> anyone got the link to the docs where they describe how to make pure python wheels and eggs my google foo is failing me

[13:42:28] <pf_moore> ronny: http://packaging.python.org/en/latest/tutorial.html#building-packaging-your-project

[13:42:38] <pf_moore> Basically setup.py bdist_wheel

[13:42:56] <pf_moore> (No idea about eggs tho, but presumably setup.py bdist_egg)

[13:48:44] <ronny> hmm

[13:48:47] <ronny> its not documented

[13:52:01] <Ivo> ronny: http://wheel.readthedocs.org/en/latest/#usage

[13:52:28] <Ivo> Your wheel is "pure" if it comes out with *-none-any.whl on the end

[13:52:58] <ronny> Ivo: im that far, now i need the same for eggs

[13:53:20] <Ivo> ronny: wha reason you want to support eggs?

[13:53:43] <ronny> Ivo: the package im making is something for install_requires, so having an egg is actually practical

[13:53:59] <ronny> eh setup_requires i mean

[13:59:25] <ronny> dstufft: so to get eggs, i'd have to do a bdist_egg on each python version?

[13:59:37] <dstufft> yup

[13:59:49] <ronny> do we have anything that automates that?

[14:00:04] <dstufft> nope

[14:00:29] <ronny> i see

[14:01:00] <Ivo> ronny: tbh just make an sdist and let setuptools generate them for setup_requires

[14:02:01] <ronny> jaraco_: i suppose there arent any plans to make this more practical?

[14:04:25] <dstufft> hopefully it won't mattch much in the future, I have a PoC that'll allow wheels to be used in setup_requires

[14:04:55] <ronny> i see

[14:05:30] <ronny> dstufft: my package hgdistver incudes tools to extrat generated version numbers from hg/git using the last reacable tag

[14:05:50] <ronny> im wondering if/how i can keep using it after packaging becomes a thing

[14:07:09] <jaraco_> ronny, no plans specifically. I'm reluctant to set up and maintain the build infrastructure that would be required to create wheels for every python/platform combination.

[14:07:22] <jaraco_> interesting you mention hgdistver. It sounds similar to setuptools_hg and hgtools.

[14:07:36] <ronny> jaraco_: it kind of is

[14:08:22] <ronny> btw, how should one pur a scm revision number into a release?

[14:08:43] <ronny> (im doing a develop release for each commit basically)

[14:08:48] <jaraco_> I asked pje about that a while ago, and the answer was that the metadata doesn't currently support it.

[14:09:11] <jaraco> So hgtools uses a hack to store the version in the 'tag' field.

[14:10:25] <ronny> hgdistver currently attaches the commit id to the revision number if its not a release tag

[14:10:55] <ronny> and im currently trying to port it to support .dev version numbering which needs to guess the next release number

[14:11:16] <jaraco> interesting. hgtools does the latter but not the former.

[14:11:34] <ronny> the later?

[14:11:47] <jaraco> adds .dev version numbering and guesses the next release number.

[14:12:15] <jaraco> The former is still ticketed - https://bitbucket.org/jaraco/hgtools/issue/15/append-changeset-hash-to-version-if-its

[14:12:55] <ronny> ah, right, its complex

[14:13:07] <ronny> hgdistver is pretty simple and single-file

[14:13:34] <jaraco> yep. every project starts out that way ;)

[14:13:57] <ronny> hmm, i completely ignored hgtools cause it does all those complex things

[14:14:11] <ronny> basically the only thing missing in hgdistver is guess next version

[14:14:44] <ronny> and for that i'll bascially require semantic versioning, and increment if the last component of the last tag is not a number

[14:17:16] <jaraco> Fair enough.

[14:17:39] <ronny> jaraco: target is single file and under 400 loc

[14:18:33] <ronny> and i might need a new name, due to git support

[14:18:54] <jaraco> hgtools has the same problem

[14:30:22] <ronny> hmm

[14:30:23] <ronny> yat

[14:30:39] <ronny> now instead of 0.1.post3 i get 0.2.dev3 :)

[14:31:27] <ronny> now all i need to test is that 0.1.1.dev as the last tag will end up as 0.1.1.dev3

[14:33:19] <ronny> jaraco: btw, got any good examples for unittesting a setuptools integration?

[15:21:08] <ronny> dstufft: btw, is there a way to upload to pypi without user/password?

[15:21:15] <dstufft> No

[15:24:00] <ronny> dstufft: any plans to support something like uploading propperly signed packages without password?

[15:24:32] <dstufft> there are plans to support something like that, but first the goal is to rewrite PyPI to be a maintinable code base

[15:24:39] <ronny> i see

[15:24:56] <ronny> howe far along is that endavor?

[15:27:04] <Alex_Gaynor> ronny: http://warehouse.python.org/

[15:36:08] <ronny> hmm, btw completely different ui?

[15:38:20] <ronny> dstufft: btw, is there a way to mark packages as unmaintained?

[15:39:07] <dstufft> not really

[15:39:26] <dstufft> you can set a classifier

[15:39:27] <dstufft> Development Status :: 7 - Inactive

[15:40:14] <ronny> i see

[15:40:30] <ronny> hmm

[15:40:51] <ronny> i dont have my pypi password anymore, lost it somewhere along the last 2 years

[15:41:21] <dstufft> what's your PyPI username

[15:43:13] <ronny> ronny

[15:55:39] <ronny> dstufft: ?

[15:56:45] <dstufft> ah you got access to it?

[16:28:38] <ronny> dstufft: openid

[16:28:46] <ronny> dstufft: but no password

[17:54:05] <dstufft> ronny: sorry went out to get some food

[17:54:13] <dstufft> ronny: there should be a password reset option, does that work ?

[19:41:33] <ronny> dstufft: i suppose i have to logout?

[19:41:44] <dstufft> might have too

[19:41:50] <dstufft> PyPI legacy is horrible

[19:49:49] <ronny> dstufft: how long does it take till a package shows up in the updated list?

[19:51:24] <dstufft> which updated list

[19:56:10] <ronny> dstufft: landing page of the package index

[19:56:31] <dstufft> on pypi.python.org? should be nearly instantly, though it may be cached for up to a minute

[19:57:00] <ronny> dstufft: i did a twine upload of hgdistver 0.17, i wont show in the list

[19:58:00] <dstufft> hm

[19:58:08] <dstufft> was there an error when you did it?

[19:58:32] <ronny> no

[19:58:36] <ronny> hmm

[19:58:51] <dstufft> https://pypi.python.org/pypi/hgdistver says it was uploaded 2013-06-27

[19:59:19] <ronny> sometihng is wron here, let me recheck

[20:00:51] <ronny> dstufft: http://bpaste.net/show/yyinVsBIW4ylKRpLhMo4/

[20:01:58] <dstufft> hrm

[20:06:28] <dstufft> ronny: hrm

[20:06:41] <dstufft> According to PyPI you've not made any changes since 2013

[20:07:47] <dstufft> ronny: can you tell me your IP address?

[20:22:26] <ronny> dstufft: moment

[20:22:52] <ronny> dstufft: currently my external ip is 188.109.215.195

[20:27:33] <ronny> omg fml

[20:29:02] <ronny> dstufft: i did something rther dumb, i orgot to push the 0.17 tag on my old harddisk before the failure, but after the release

[20:29:36] <dstufft> ronny: doesn't look like PyPI has any POST requests from you with the requests user agent (what twine uses)

[20:29:42] <dstufft> oh wait

[20:29:45] <dstufft> ernest just found one

[20:29:47] <dstufft> or some

[20:30:02] <dstufft> ronny: does that mean the issue is on your side?

[20:30:13] <ronny> hmm

[20:30:17] <ronny> im confused

[20:30:55] <ronny> dstufft: http://bpaste.net/show/vnWUHg9igoeONufB3crq/ - no result on the website

[20:31:58] <ronny> dstufft: ok, pypi is a mess, when using pypi.python.org/pypi instead of what i used before, it works

[20:32:28] <dstufft> ah ha

[20:32:29] <dstufft> yes

[20:32:37] <dstufft> pypi.python.org/ just redirects

[20:33:28] <ronny> that like totally fscked

[20:33:36] <ronny> the tools report succesa and nothing happened

[20:33:54] <ronny> hmm

[20:34:14] <ronny> and i was like i uploaded version 0.17, and it landed, what the heck is wrong here ^^

[20:34:21] <ronny> now 0.18 has correctly landed

[20:36:00] <dstufft> yea sorry

[20:36:03] <dstufft> old pypi is real bad

[20:36:06] <dstufft> new pypi will be better

[20:38:30] <EWDurbin> i concur!

[20:40:04] <ronny> :)

[20:40:14] <ronny> dstufft: when will it be deployed?

[20:41:09] <dstufft> dunno! I'm working on it at the moment, but it's not close enough to feature parity to be deployed yet

[20:41:19] <dstufft> I know richard is planning on sprinting at PyCon, or I think he is anyways

[23:40:43] <agronholm> richard who

[23:40:57] <agronholm> the new pypi is something I would like to sprint on too

[23:47:05] <dstufft> agronholm: Richard Jones :]

[23:47:38] <dstufft> agronholm: I'm trying to get https://github.com/pypa/warehouse/pull/248 done before the PyCon sprints! It unblocks a bunch of work for implementing the logged in user views :]

Log file Viewer

Help | Karma | Search:

#pypa logs for Saturday the 5th of April, 2014