[11:01:30] <mgedmin> it would be nice if https://status.python.org/ said where to report outages
[12:21:45] <mkoskar> Hello, how does it happen that I can find a package at <https://pypi.python.org/simple> but not with search at <https://pypi.python.org/pypi>? That actually means that e.g., 'pip install ...' will install something 'pip search ...' doesn't show.
[12:23:21] <mgedmin> releases can be hidden from view
[12:23:36] <mgedmin> pip finds them (via /simple), users don't see them
[12:24:01] <mgedmin> iirc when all releases of a project are hidden, /pypi/projectname returns 404 (e.g. pygame)
[12:24:49] <mgedmin> when there's more than one non-hidden release /pypi/projectname shows a table letting the user pick the one they want
[12:25:04] <mgedmin> when there's exactly one non-hidden release, /pypi/projectname shows it
[12:27:30] <mkoskar> mgedmin: I see. Hmm, it's pretty confusing I think. I stumbled over it when trying to install 'pyenv'. I think it would make sense to have 'pip install' and 'pip search' consistent.
[12:28:55] <mgedmin> imagine my surprise when I first went to /pypi/pygame and saw it redirect me to /pypi/Pygame, which reports 404
[12:29:10] <mgedmin> (/pypi/pygame/1.8.0 shows that release like everything's fine)
[12:29:36] <mgedmin> (of course there are no downloadable pygame files on pypi anyway, and the pygame maintainers seem to be MIA)
[12:33:50] <mkoskar> mgedmin: yes that is very odd, indeed
[12:34:14] <mgedmin> I'm sure warehouse (the upcoming rewrite of pypi) will fix everything ;)
[12:35:48] <mgedmin> e.g. https://pypi.python.org/pypi?%3Aaction=search&term=pygame&submit=search ?
[12:35:50] <mkoskar> well, search through web interface, and also 'pip search'
[12:36:00] <mgedmin> hm, I don't know how it works under the hood actually
[12:36:16] <mgedmin> there were some complaints on the distutils-sig@ mailing list about search failing to return results
[12:36:23] <mgedmin> and it wasn't related to hidden releases
[12:36:39] <mgedmin> something something search index got wedged or something? but I thought admins got it unwedged?
[12:39:14] <mkoskar> mgedmin: yes, that might be the case, I dunno, I'm relatively infrequent pip and pypi user
[12:40:24] <mkoskar> ok, I might have a look at this particular package later down the line, and if it persists I will raise it on ML, maybe it's a known issue maybe not
[20:34:40] <nedbat> i am having difficulties where packages are installed twice, once as -e (in the venv/src directory, with a .egg-link file pointing to it), and once as a real installation (a .egg-info file).
[20:35:13] <nedbat> python imports the src version, but pip looks at the .egg-info version when deciding whether to upgrade. As a result, I am stuck on the old version.
[20:35:30] <nedbat> is there a way to invoke pip install so that this won't happen? (other than completely re-installing everything every time)
[21:05:52] <jwhite007> i'm attempting a pip install of pandas and get "ImportError: /my/path/libblas.so: undefined symbol: _gfortran_copy_string" any suggestions?
[21:25:48] <nedbat> jwhite007: make sure you have fortran install on your machine
[22:23:48] <jwhite007> nedbat: gfortran is in my path.
[22:24:22] <jwhite007> nedbat: i think i have the wrong libblas.so