pmxbot IRC Log Viewer

[14:23:06] <sigmavirus24> dstufft: ping

[14:25:46] <sigmavirus24> Or perhaps others would know where I'd look if I wanted to see what code handles package uploads to PyPI

[16:10:49] <dstufft> sigmavirus24: um, I think it's the file_upload function

[16:10:53] <dstufft> I think tha'ts what it's called

[16:10:55] <dstufft> in webui.py

[16:11:09] <sigmavirus24> aha thanks

[16:11:33] <dstufft> https://bitbucket.org/pypa/pypi/src/1bab886066b17880ada4cec987d3a12112e47d2a/webui.py?at=default&fileviewer=file-view-default#webui.py-2734

[16:12:02] <dstufft> in general, all of our view functions are in webui.py and all of our data manipulation functions are in store.py

[16:16:42] <sigmavirus24> I'm just trying to figure out where it rejects an upload as being a dupe

[16:18:02] <sigmavirus24> Aha that's https://bitbucket.org/pypa/pypi/src/1bab886066b17880ada4cec987d3a12112e47d2a/webui.py?at=default&fileviewer=file-view-default#webui.py-2834

[16:18:39] <sigmavirus24> The funny thing is that the exception message is returned as the "reason" for the response

[16:19:15] <sigmavirus24> e.g., "HTTP/1.1 400 A file named "foo-1.2.3.tar.gz" already exists for foo-1.2.3. To fix p roblems...."

[16:19:40] <dstufft> yea

[16:19:48] <dstufft> hysterical raisins

[16:20:04] <dstufft> Richard wanted to add a header IIRC that returned the reason

[16:20:13] <dstufft> and people were like "just use the reason, that's what it's there for!!"

[16:20:29] <dstufft> now it's impossible to communicate anything to use except via a 400 response's reason

[16:48:03] <sigmavirus24> dstufft: that's amazing

[16:48:06] <dstufft> sigmavirus24: how hey, did you add a twine register?

[16:48:11] <dstufft> oh hey*

[16:48:18] <sigmavirus24> yeah people have been bugging me offline about it

[16:48:23] <dstufft> cool

[16:48:27] <sigmavirus24> I know warehouse won't need it

[16:48:30] <sigmavirus24> But yolo

[16:48:32] <dstufft> I think travis could use it

[16:48:36] <sigmavirus24> travis-ci?

[16:48:39] <dstufft> yea

[16:48:42] <sigmavirus24> hah

[16:48:44] <sigmavirus24> awesome

[16:48:49] <sigmavirus24> I also refactored things to make stuff ... ncier

[16:48:49] <dstufft> jezdez: ^^^

[16:48:50] <sigmavirus24> *nicer

[16:49:16] <dstufft> jezdez: was making PRs to travis to switch them from setup.py upload to twine, but the lack of register was an issue IIRC

[16:49:18] <sigmavirus24> I'm pretty happy with https://github.com/pypa/twine/pull/121, thoughts?

[16:49:23] <sigmavirus24> dstufft: aha

[16:49:37] <sigmavirus24> I want to fix the PKGINFO warnings

[16:50:10] <dstufft> sigmavirus24: looks good

[16:50:13] <sigmavirus24> I was considering a flag for upload to register on failure

[16:50:24] <sigmavirus24> not sure I want to add more flags to upload

[16:50:32] <dstufft> sigmavirus24: you should make sure I didn't change that message in warehouse

[16:50:40] <sigmavirus24> lol

[16:50:42] <dstufft> sigmavirus24: I think a "register on failure" flag is kinda silly

[16:50:44] <sigmavirus24> well I'm not testing the whole thing

[16:50:49] <sigmavirus24> but yeah

[16:50:51] <sigmavirus24> I'll check

[16:51:06] <dstufft> because you literally only need to register once in a project's lifetime

[16:51:15] <sigmavirus24> yep

[16:51:54] <dstufft> sigmavirus24: I'm happy to make the exact error message guarenteed to be stable, we'll just need to add a comment to say that

[16:52:04] <sigmavirus24> You did change it in legacy at least: https://github.com/pypa/warehouse/blob/master/warehouse/legacy/api/pypi.py#L597

[16:52:10] <sigmavirus24> I'm not concerned

[16:52:25] <sigmavirus24> I can also future-proof that for warehouse

[16:52:41] <sigmavirus24> Actually also maybe https://github.com/pypa/warehouse/blob/master/warehouse/legacy/api/pypi.py#L607

[16:52:48] <sigmavirus24> lol

[16:53:03] <sigmavirus24> Also can I start sending sha256 checksums too?

[16:53:16] <sigmavirus24> Or should I? (e.g., more future proofing)

[16:54:06] <dstufft> it wouldn't hurt to send it in addition to the md5 hash

[16:54:16] <dstufft> opena ticket on warehouse to support it

[16:54:28] <sigmavirus24> Roger wilco

[16:54:47] <sigmavirus24> I remember PyPI was going to move that direction and was waiting on devpi which supports it already

[16:54:58] <dstufft> yea

[16:55:18] <dstufft> that'll probably happen shortly after warehouse goes live

[16:55:41] <dstufft> I'm not doing too much additional evolution in PyPI until we lose the albatross

[16:57:26] <sigmavirus24> Makes sense

[16:57:43] <sigmavirus24> dstufft: I might be interested in sending a PR to do that because I already know where the MD5 validation is, but I probably won't have time

[16:59:18] <dstufft> I wish there was a less ugly name for CoC than CODE_OF_CONDUCT.md

[17:07:18] <sigmavirus24> lol

[17:07:21] <sigmavirus24> CONTRIBUTOR_COVENANT.md

[17:07:45] <sigmavirus24> I use github.com/coralineada/contributor_covenant so

[17:07:48] <sigmavirus24> makes sense to me

[17:10:28] <willingc> dstufft: Heads up on 3.5 and scipy/numpy/pandas and pip https://bugs.python.org/issue25088

[17:10:43] <willingc> dstufft: likely a 3.5 issue

[17:11:28] <dstufft> oh godoy

[17:11:34] <willingc> lol

[17:11:50] <dstufft> does setup.py install work

[17:11:54] <dstufft> without pip

[17:12:13] <willingc> on a happy note, lots of positive pr at djangocon on your work :)

[17:12:23] <willingc> haven't tried setup.py

[17:12:39] <dstufft> that'll tell us if it's pip's fault or setup.py's fault

[17:13:01] <dstufft> given flask works, and scipy and co do a lot of customization of setup.py, I'm guiessing it's their fault

[17:13:05] <dstufft> but that's just a guess

[17:13:12] <willingc> what exact command should I execute

[17:13:47] <sigmavirus24> willingc: I would try first: "python setup.py bdist_wheel"

[17:13:57] <sigmavirus24> that's what newish pips are doing before installing but after download

[17:14:04] <willingc> ok

[17:15:09] <willingc> sigmavirus24: /Library/Frameworks/Python.framework/Versions/3.5/Resources/Python.app/Contents/MacOS/Python: can't open file 'setup.py': [Errno 2] No such file or directory

[17:15:54] <dstufft> willingc: you need to download the scipy file from PyPI and unpack it

[17:15:58] <dstufft> setup.py will be inside there

[17:16:07] <willingc> ahh...will do

[17:18:39] <dstufft> https://www.pypa.io/en/latest/code-of-conduct/

[17:22:40] <sigmavirus24> dstufft: A+

[17:22:50] <willingc> dstufft, sigmavirus24: scipy depends on numpy so tried with numpy. Downloaded source, unpacked, and ./setup.py with these results

[17:22:57] <willingc> Running from numpy source directory.

[17:22:57] <willingc> warnings.warn(msg)

[17:23:38] <willingc> dstufft: /Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/distutils/dist.py:261: UserWarning: Unknown distribution option: 'test_suite'

[17:23:54] <sigmavirus24> hm

[17:24:18] <willingc> Nicely done on CoC

[17:24:50] <sigmavirus24> willingc: can you do `python -c 'import setuptools; print(setuptools.__version__)'`?

[17:25:24] <willingc> 18.2

[17:25:54] <dstufft> the test_suite thing should be able to be safely ignored

[17:26:01] <dstufft> did you do ``setup.py`` or ``setup.py install``

[17:26:08] <willingc> setup.py

[17:27:53] <dstufft> willingc: try it with setup.py install

[17:28:38] <willingc> looks like that is doing something

[17:30:55] <willingc> dstufft: Final statement Copying build/temp.macosx-10.6-intel-3.5/libnpymath.a -> /Users/carol/Projects/myscipy/lib/python3.5/site-packages/numpy/core/lib

[17:31:23] <dstufft> willingc: and ``pip install numpy`` fails as it did in the ticket?

[17:32:13] <willingc> Actually now pip is saying requirement already satisfied.

[17:32:38] <willingc> yet the interpreter doesn't import numpy without error

[17:34:38] <dstufft> lol

[17:34:44] <dstufft> I'll take a poke at it in a bit

[17:34:56] <dstufft> I have afeeling it's gonna be gnarly :/

[17:35:23] <willingc> Ok. I'm driving to CalPoly later. Will be meeting up with some Jupyter folks tomorrow. I'm hoping it is not.

[17:35:43] <willingc> thanks dstufft and sigmavirus24

[17:36:31] <sigmavirus24> Drive safely willingc :)

[17:36:39] <willingc> Will do :)

[17:46:19] <dstufft> so many repositories

[17:47:30] <sigmavirus24> My former coworkers who are type/haskell fiends just gave me a terrible idea: A flake8 plugin for python 3.5 only that integrates with mypy/typing module

[17:47:58] <dstufft> why is that a terrible idea

[17:48:03] <dstufft> sounds like exactly waht typing is for

[17:48:31] <sigmavirus24> I'm concerned about the sudden increase in hatemail I'll receive

[17:53:27] <dstufft> lol

[17:54:05] <sigmavirus24> "Stop encouraging people makign their python easier to reason about"

[18:29:29] <willingc> dstufft: Ok. I think that I figured out part of the scipy install issue. I did a ``pip show --files numpy`` and received a `Cannot locate installed-files.txt`. I was able to pip install pandas. I finally got the idea after another ``pip install scipy`` attempt that perhaps gfortran wasn't being found when ``pip install`` numpy or scipy and it was hanging with no indication. That seemed to have been the problem. I installed gfortran for os x using

[18:29:50] <willingc> http://hpc.sourceforge.net/

[18:30:00] <willingc> ``pip install scipy -v`` then succeeded after 8min 39s. Imports worked with no errors too.

[18:30:01] <willingc> I love when things fail with no indication of why :D

[18:41:42] <sigmavirus24> willingc: silent failure is better than boistrous failure =P

[18:42:01] <willingc> sigmavirus24: lol

[18:57:16] <pmxbot> jaraco pushed 1 commit to setuptools (https://bitbucket.org/pypa/setuptools/) :

[18:57:16] <pmxbot> Add the PyPA Code of Conduct

[19:26:17] <qwcode> dstufft, I converted the pip docs to htmldir. the direct rtd link is working fine. http://pip.readthedocs.org/en/stable/installing/, but the pypa.io link is not https://pip.pypa.io/en/stable/installing.html

[19:27:04] <qwcode> dstufft, if you can't fix soon, I'll probably switch back

[19:28:20] <dstufft> qwcode: you probably need to purge

[19:28:23] <dstufft> sec

[19:29:49] <dstufft> qwcode: try now

[19:30:58] <qwcode> I would expect this to redirect properyl, but it doesn't https://pip.pypa.io/en/stable/installing.html

[19:31:53] <dstufft> oh right

[19:32:00] <dstufft> that's a RTD bug

[19:32:08] <dstufft> Might be able to fix it in Fastly though

[19:32:17] <qwcode> but it doesn't redirect using the direct rtd link either http://pip.readthedocs.org/en/stable/installing.html

[19:32:26] <qwcode> yea, agreed, rtd problem

[19:32:53] <dstufft> I think a Fastly rule to redirect $WAHTEVER.html to $WHATEVER/ should solve it

[19:32:56] <qwcode> they supposedly support that. I added the redirect in our admin console

[19:32:58] <dstufft> I can do that

[19:33:12] <dstufft> I don't care about redirects on the pip.rtd link, do you?

[19:33:20] <qwcode> I chatted in #readthedocs, but nothing so far

[19:34:04] <qwcode> no, I don't care, I just thought maybe it would bubble up to your fastly layer

[19:34:34] <dstufft> it would if it was working there yea

[19:34:58] <dstufft> but we can paper over it in Fastly

[19:35:36] <qwcode> ok

[19:36:20] <qwcode> since youre add it, can you do it for packaging.python.org too? and I'll convert that as well?

[19:36:29] <dstufft> yea

[19:39:23] <qwcode> ok, PyPUG is converted and rebuilt

[19:48:33] <dstufft> qwcode: https://pip.pypa.io/en/stable/installing.html should work now

[19:49:22] <sigmavirus24> dstufft: thoughts on moving pkginfo into PyPA? (https://bugs.launchpad.net/pkginfo/+bug/1437570/comments/1)

[19:54:05] <qwcode> dstufft, yea, it does. nested dirs with index.rst pages are still a problem though https://pip.pypa.io/en/latest/reference/index.html

[19:54:35] <dstufft> sigmavirus24: no real opinion, seems reasonable

[19:54:41] <dstufft> qwcode: bleh, OK I'll take a look

[19:55:06] <dstufft> hm

[19:55:15] <dstufft> ah

[19:55:16] <dstufft> I see

[19:55:24] <dstufft> it's because index.html needs more chopped off

[19:55:31] <dstufft> than just the .html

[19:57:28] <dstufft> qwcode: packaging.python.org should be working now

[19:57:33] <dstufft> going back to pip to fix that

[19:59:11] <sigmavirus24> dstufft: yeah, it's a matter of "pkginfo" looks unmaintained (hasn't seen much development since 2013) so maybe the people that use it (packaging people) can take care of it until it's irrelevant

[20:02:02] <qwcode> dstufft, great, thanks

[20:03:54] <dstufft> qwcode: index.rst urls should work on pip now

[20:04:01] <qwcode> dstufft, great

[20:07:43] <dstufft> qwcode: looks good, thanks forg etting that done :)

[20:12:19] <dstufft> gonna lay down for a bit

[21:52:28] <xafer> hello, I was wondering what is the point of having "[metadata] requires-dist" in https://github.com/pypa/twine/blob/master/setup.cfg ?

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Sunday the 13th of September, 2015