pmxbot IRC Log Viewer

[07:59:37] <mgedmin> 503 error from pypi! time to clean my cookies! WHY IS THIS HAPPENING?

[16:56:39] <tos9> Having the "setup script specifies an absolute path" problem again.

[16:56:48] <tos9> Has anyone actually been able to figure out what's up there

[16:58:20] <danielwholth> why would it do that

[16:59:15] <tos9> Why would what do what?

[17:33:38] <cerivera> with my custom package, when i do pip install —upgrade PACKAGE. it tells me it's downloading version 0.12.6 (latest). i verify the version after install in ipython using PACKAGE.__version__

[17:33:45] <cerivera> however, when i do pip freeze | grep PACKAGE

[17:33:50] <cerivera> it shows an older package version

[17:34:34] <cerivera> basically, pip install —upgrade always downloads and installs 0.12.6 because it thinks an older version is installed

[18:10:43] <nedbat> there used to be download stats on the PyPI pages, but I don't see them now. Are they gone, or just harder to find?

[18:11:38] <dstufft> nedbat: which download stats? there are ones that say the daily/week/ymonthly download numbers is that what you mean?

[18:12:13] <nedbat> dstufft: I want to know how many people are downloading the 3.2 kit for coverage.py compared to the 3.3 and 3.4

[18:12:56] <dstufft> nedbat: you mean the per file download counts?

[18:13:03] <nedbat> dstufft: yes

[18:15:17] <dstufft> nedbat: yea that's not in the UI anymore, it's in the API still though

[18:15:51] <dstufft> https://pypi.python.org/pypi/coverage/json

[18:16:20] <nedbat> dstufft: thanks, that's UI enough for me! :)

[18:21:20] <Alex_Gaynor> nedbat: We're also working on getting more detailed download stats into a databass that can be queries reasonably

[18:22:07] <nedbat> Alex_Gaynor: great to hear. the json did the job for me! :)

[19:05:01] <ionelmc> god damn, why would pip segfault?

[19:05:25] <ionelmc> https://www.irccloud.com/pastebin/DtOQwnQi

[19:06:05] <ionelmc> segfaults in Py_Finalize (from Py_Exit)

[19:27:22] <ionelmc> anyone using appveyor for building wheels?

[19:27:40] <ionelmc> what's good way to auth the upload to pypi ?

[19:27:44] <ionelmc> dstufft: ping

[19:28:00] <dstufft> hmm?

[19:29:16] <josePhoenix> There's a form to upload docs on testpypi... where does it go? Same as real pypi?

[19:30:46] <dstufft> http://test.pythonhosted.org/ "?

[19:31:03] <josePhoenix> ahh :)

[19:31:16] <josePhoenix> that information is hard to discover

[19:31:55] <dstufft> I more or less recommend RTD not upoading docs to pypi

[19:32:12] <josePhoenix> RTD doesn't work for our docs

[19:32:43] <josePhoenix> I talked to someone about it and they basically said "yeah, we'd like to support scientific python packages better, eventually"

[19:32:49] <ionelmc> dstufft: well i guess i can just take the wheels (they'd be artifacts) and upload them myself

[19:33:52] <ionelmc> dstufft: is this really needed https://raw.githubusercontent.com/pypa/python-packaging-user-guide/master/source/code/run_with_compiler.cmd ?

[19:34:11] <ionelmc> it seemed to compile just fine on 3.4 on appveyor (without that)

[19:34:13] <dstufft> I know very little about Windows

[19:36:04] <ionelmc> pf_moore: ping

[20:34:56] <pf_moore> ionelmc: pong

[20:38:26] <pf_moore> ionelmc: if it's about that run_with_compiler script, it's needed for C extensions. If you don't have them, you don't need it (but you also don't need to build the wheels on Appveyor)

[20:40:58] <pf_moore> It also isn't needed if you only do 32-bit builds, as VC express 2010 includes the 32-bit compilers

[20:41:42] <pf_moore> But it is needed for 64-bit, maybe it's a bit complicated but it's known to work so why mess with it?

[20:41:59] <ionelmc> oh well

[20:42:02] <ionelmc> troublemaking :)

[20:42:15] <pf_moore> :-)

[20:42:35] <ionelmc> tho i wonder about the sample PS scripts

[20:42:49] <ionelmc> appveyor allready has all the pytohns

[20:42:55] <pf_moore> You said "it seemed to compile just fine" - what did, and what versions were you building?

[20:42:57] <ionelmc> pf_moore: is there something wrong with them?

[20:43:05] <ionelmc> pf_moore: 32bit apparently

[20:43:09] <pf_moore> which scripts?

[20:43:24] <ionelmc> scripts?!

[20:43:35] <pf_moore> "i wonder about the sample PS scripts"

[20:43:43] <ionelmc> https://github.com/pypa/python-packaging-user-guide/blob/master/source/code/install.ps1

[20:44:13] <pf_moore> Installs pip, wheel and any Pythons that aren't already present.

[20:44:21] <ionelmc> i mean, it's just a minor version difference

[20:44:27] <pf_moore> When I set it up, Appveyor didn't have all the Python versions

[20:44:43] <ionelmc> appveyor has these:

[20:44:46] <ionelmc> 2.7.8 x86 (C:\Python27 - default in PATH)

[20:44:46] <ionelmc> 2.7.8 x64 (C:\Python27-x64)

[20:44:46] <ionelmc> 3.3.5 x86 (C:\Python33)

[20:44:46] <ionelmc> 3.3.5 x64 (C:\Python33-x64)

[20:44:46] <ionelmc> 3.4.1 x86 (C:\Python34)

[20:44:47] <ionelmc> 3.4.1 x64 (C:\Python34-x64)

[20:44:57] <ionelmc> ah

[20:45:13] <ionelmc> well yeah,

[20:45:15] <pf_moore> So if you use those the PS script won't download them. No problem.

[20:45:23] <ionelmc> i was thinking that it;s better to have faster builds

[20:45:29] <pf_moore> But if you want 2.7.9, it'll do it for you

[20:45:29] <ionelmc> aaaah

[20:45:31] <ionelmc> got it

[20:46:00] <ionelmc> pf_moore: will the ps script make the last install the "default" python ?

[20:46:34] <pf_moore> Honestly, I don't know. You should specify the full path to avoid uncertainty

[20:47:52] <pf_moore> The silent MSI install is not very well documented and I didn't have the time or inclination to experiment just so users could take a risk doing "python" without a path...

[20:49:20] <pf_moore> BTW, the script also avoids trying to install pip if it's present (3.4, 2.7.9). So it doesn't do more work than it needs, shouldn't impact build speed much.

[20:49:51] <pf_moore> If setting up the env is taking up too much of the build time, write more tests ;-)

[20:51:11] <ionelmc> heh

[21:21:52] <tdsmith> when i pass --install-data to setup.py install, why is distlib trying to create a path in /Library in install_data :(

[21:23:58] <tdsmith> distutils*

[21:24:20] <tdsmith> at least i assume it's distutils' fault since i'm passing --single-version-externally-managed

[21:38:14] <tdsmith> ohhhh no, translate-toolkit isn't expecting the apple system framework build layout

[21:38:22] <tdsmith> and it's confused

Log file Viewer

Help | Karma | Search:

#pypa logs for Wednesday the 10th of December, 2014