Log file Viewer

[01:48:02] <guitar3> dstufft: 67M for the manylinux1 wheels, the os x wheels are ~57M and I can upload those without a problem.

[02:18:35] <warner> hey, I'm getting an unexpected error while doing "twine register" on a new sdist tarball: has anyone see "HTTPError: 410 Client Error: This API is no longer supported, instead simply upload the file. for url: https://upload.pypi.io/legacy/" before?

[02:18:52] <warner> this is with twine-1.7.4

[02:24:19] <warner> downgrading to twine-1.6.5 let me upload the package successfully

[02:24:42] <warner> guess I'll file a ticket

[04:34:43] <[Tritium]> guitar3, dstufft: odd that 67M is a problem... pyqt5 has 3 wheels much larger than that on pypi

[08:32:35] <joke2k> hi, I've a problem with virtualenvwrapper, can someone help me?

[08:33:13] <apollo13> just ask your question

[08:35:46] <joke2k> apollo13, I've a fresh ubuntu 16 with ohmyzsh, when I try to use `mkvirtualenv test` the command does not create a folder in my $WORKON_HOME

[08:36:08] <apollo13> joke2k: does it work in bash?

[08:36:29] <joke2k> if I reload with `source /usr/local/bin/virtualenvwrapper.sh` everything works

[08:36:47] <apollo13> maybe a zsh & venvwrapper issue?

[08:36:58] <apollo13> I can say it works for me at least

[08:37:10] <apollo13> how are you loading venvwrapper initially?

[08:37:28] <apollo13> I have source ~/.local/bin/virtualenvwrapper.sh in my .zshrc

[08:39:13] <joke2k> I've a `~/.oh-my-zsh/custom/profile.zsh` that loads ~/.myprofile

[08:39:41] <apollo13> joke2k: mhm, that would spawn a new process and not include those?

[08:40:14] <joke2k> probably

[08:40:40] <joke2k> i know that is a insane configuration but I don't like to add stuff into ~/.*rc files

[08:41:09] <joke2k> i try to move into .zshrc

[08:44:24] <joke2k> it works

[08:44:27] <apollo13> joke2k: well, but if you source that in a subshell it won't be in your parent process

[08:44:36] <apollo13> I am surprised that exporting env vars or anything works

[08:44:43] <joke2k> me too

[08:44:45] <apollo13> what happens if you source ~/.oh-my-zsh/custom/profile.zsh or . ~/.oh-my-zsh/custom/profile.zsh

[08:44:57] <apollo13> sorry my shell foo is not strong enough to be of real help

[08:46:26] <joke2k> apollo13, thank you :D

[08:52:18] <wiggy> is there some way to see why pip things something "it is not compatible with this Python"

[16:50:24] <tdsmith> given an unpacked sdist *or* scm repository can i learn what the name of the python package is?

[16:50:36] <tdsmith> i think i should be able to run setup.py egg_info, at least

[16:50:44] <tdsmith> is this a function that pip could or should have?

[17:05:23] <dstufft> tdsmith: python setup.py egg_info yea, or uh setup.py --name

[17:08:34] <tdsmith> oh that's handy

[17:49:48] <margot> when I run 'pip download -d /tmp/ tables' it downloads to /tmp/ just fine, but then it tries to run 'python setup.py egg_info' is there any way to prevent that from happening?

[17:50:37] <margot> this is with pip version 8.1.2

[18:01:51] <dstufft> margot: it needs to run setup.py egg_info to detemine the dependencies

[18:04:44] <margot> dstufft: thanks for the reply, I don't want it check for dependencies though, I just want to it to download the package and not error out midway through the requirements.txt file

[18:05:37] <margot> the problem is that it's looking for a "dependency" that's a header file of a library I want to avoid installing on the specific machine

[18:06:00] <margot> I just want to package all the requirements on that machine, not actually run anything

[18:06:58] <margot> I can certainly do some bash shenanigans and I guess make sure that the tables requirement is at the end of the requirements.txt file, but that seems sort of trashy...

[18:18:58] <margot> I would think that the --no-deps option would prevent this, but it doesn't

[20:17:06] <nedbat> anyone have advice about pypa/pip issue #3862 ?

[20:17:57] <nedbat> https://github.com/pypa/pip/issues/3862 seems like we can't move from 8.1.1 to 8.1.2?

[20:18:16] <ngoldbaum> that issue says you can do "pip install -U pip", though

[20:18:32] <ngoldbaum> it's just when you specify -u pip==8.1.2 that it dies

[20:19:53] <nedbat> ngoldbaum: hmm, turns out we can install latest, and then install pinned to 8.1.2.

[20:20:07] <nedbat> ngoldbaum: we didn't see a way to go to specifically 8.1.2, but this is a workaround.

[20:31:18] <tdsmith> this succeeds for me on os x: mktmpenv; pip install pip==8.1.1; pip install -U pip==8.1.2

[20:31:21] <tdsmith> suspicious

[20:41:16] <dstufft> nedbat: I suspect it's due to Debian's patching.

[20:41:30] <dstufft> -> /usr/lib/python2.7/dist-packages/pip/req/req_install.py

[20:41:36] <dstufft> that is a Debian path

[20:41:51] <ngoldbaum> they had enough time to patch pip, but not enough to ensure that the patches were actually correct :/

[20:42:55] <nedbat> dstufft: hmm, the plot thickens!

[20:44:11] <dstufft> why it happens with ==8.1.2 but not without, I haven't the slightest idea

[20:44:41] <dstufft> but it's not hardly the first really weird error that comes from the patches

[20:45:10] <nedbat> dstufft: any reason to think the two-step dance of "update to latest", "update to 8.1.2" would be bad in the future?

[20:47:09] <dstufft> nedbat: I'm not sure what you're atempting to do here, are you just wanting to make sure that you have 8.1.2 installed? or do you actually want the latest?

[20:47:25] <nedbat> dstufft: we want to pin pip, so ==8.1.2

[20:47:45] <dstufft> nedbat: could try python get-pip.py pip==8.1.2

[20:48:48] <nedbat> dstufft: we're going with the two-step for now I think

[20:48:56] <nedbat> i appreciate the discussion :)

[20:48:57] <dstufft> okay

[20:50:47] <apollo13> python-pip is never getting on my system anymore, nothing but issues :/

[20:51:00] <apollo13> always install via get-pip

[20:51:06] <ngoldbaum> https://twitter.com/durin42/status/716752300749037570

[20:57:08] <apollo13> I am on centos for my servers, not sure I wanna look what the version would be there if they had xscreensaver^^

[20:59:39] <ngoldbaum> i believe augie's opinion there is also informed by the time the debian mercurial maintainer noticed tests were failing in their pacakage and just disabled the tests

[20:59:45] <ngoldbaum> or the time they broke TLS in pip...

[21:03:46] <apollo13> ngoldbaum: well, even more fun, my pip list --outdated wouldn't list packages since they debundled everything and now you oddly enough would get different Version classes which no longer compared sensibly :/

[21:19:25] <dstufft> apollo13: oh yea, that one was because they debundled pip, but didn't debundle setuptools

[21:19:47] <apollo13> hehe, never dug down to the bottom of it

[21:19:54] <apollo13> I just said screw it and threw it over board

[21:29:09] <ceridwen> I have an awkward problem with setuptools that I don't know how to solve. The brief version: I have some code from another project I'm using. This code is complicated and fragile, partly just because it's solving a collection of awkward problems. It's set up as an outer shell script that calls a set of inner shell scripts, some of which import other files in the project, glued together with sys.path ha

[21:29:09] <ceridwen> cking. I want to distribute this code (with some new code that uses it) with setuptools without having to completely rework the inherited code.

[21:38:54] <ceridwen> The directory layouts of the project with the code I'm using first and then my project second: https://paste.debian.net/783969/

[21:46:31] <ceridwen> All those files in virt/ are Python. I could rename them with .py so that setuptools will install them, but then their executable bits won't be set.

[21:47:32] <ceridwen> The files in virt/ need to import adtlog.py and VirtSubproc.py from lib/.

[21:48:24] <ceridwen> The files in virt/ also don't have anything like a main entry point function.

[21:49:59] <ceridwen> The two standard ways of doing this are with 'console_scripts' and 'scripts'. I don't know how to get 'scripts' set up so they can import the files they need to import from lib/, and since they're not nicely-structured as functions, I don't know how to get 'console_scripts' to work.

[21:50:02] <ceridwen> Does anyone have any suggestions?

[22:00:28] <guitar3> dstufft: just following up on pystan's pypi limit. Might it be boosted to 70M? Still hitting the "Client error: File too large" error when using twine.

[22:00:44] <dstufft> guitar3: ah yea, moment

[22:00:56] <guitar3> dstufft: thank you

[22:03:16] <dstufft> guitar3: ok, should be set to 75MB now

[22:06:30] <guitar3> dstufft: it's working. thank again

[22:06:46] <dstufft> guitar3: no problem!