PMXBOT Log file Viewer

Help | Karma | Search:

#pypa logs for Sunday the 14th of February, 2016

(Back to #pypa overview) (Back to channel listing) (Animate logs)
[19:40:56] <paroneayea> hello, *
[19:42:35] <paroneayea> I was astounded to see that "setup.py bdist_egg" actually *transformed* one of the python files I'm working with
[19:42:41] <paroneayea> the original source file had:
[19:42:51] <paroneayea> except self._encode_error, exc:
[19:43:01] <paroneayea> the one that came out the other end was:
[19:43:03] <paroneayea> except self._encode_error as exc:
[19:43:11] <paroneayea> whoa where did this happen? asd is this documented?
[19:43:14] <paroneayea> this was for anyjson
[19:43:53] <paroneayea> a number of other changes too...
[19:44:16] <nanonyme> eggs are magic :(
[19:44:33] <paroneayea> oh
[19:44:36] <paroneayea> I bet it's this:
[19:44:41] <paroneayea> if sys.version_info >= (3, 0):
[19:44:41] <paroneayea> extra.update(use_2to3=True)
[19:44:43] <paroneayea> in the setup.py
[19:44:45] <paroneayea> shit!
[19:44:48] <nanonyme> Also the latter is silly
[19:45:09] <nanonyme> Unless you're supporting prehistoric Python2 versions
[19:45:16] <nanonyme> (I think 2.5 or so?)
[19:45:26] <paroneayea> nanonyme: well it's a pretty outdated package too, unfortunately
[19:45:28] <nanonyme> I mean former is silly
[19:46:08] <nanonyme> paroneayea, well, the package would not run on your machine *at all* without the change
[19:46:16] <nanonyme> Or rather, in your Python interpreter
[19:46:28] <paroneayea> nanonyme: right, I know
[19:46:40] <paroneayea> I'm just surprised that it was done in this automated way
[19:46:49] <paroneayea> I'm packaging mediagoblin's dependencies for guix, and this was a dep of kombu
[19:46:55] <nanonyme> Right
[19:47:01] <nanonyme> So why can't you use wheels?
[19:47:05] <paroneayea> and like, whoa, why is the tarball I've pulled down not working right on py3.4, it's working here..
[19:47:15] <nanonyme> Heh
[19:47:43] <paroneayea> nanonyme: honestly I know very little about wheels, but anyway the guix process kind of doesn't do either
[19:47:58] <paroneayea> nanonyme: but I was *comparing* to an egg I had pulled down into a virtualenv
[19:48:08] <paroneayea> to figure out why it worked there and not with my system one
[19:49:00] <nanonyme> Out of curiosity, what exactly is calling the 2to3 transformation?
[19:49:16] <paroneayea> nanonyme: the setup.py itself, it appears!
[19:49:25] <paroneayea> which is how it gets transformed in the bdist_eg
[19:49:26] <paroneayea> egg
[19:49:28] <paroneayea> very strange
[19:49:50] <nanonyme> That's insane. Why would the guy not have ran 2to3 *first, then committed the code into his repo and released it
[19:50:35] <nanonyme> Or just made bloody sure his code ran on both...
[19:50:49] <paroneayea> the code also hasn't been updated for a few years
[19:50:58] <paroneayea> anyway I'm just the messenger er... packager
[19:51:11] <nanonyme> Does the code have a maintainer?
[19:51:41] <paroneayea> not an active one
[19:51:43] <nanonyme> Or rather, does it have a maintainer upstream? If it does, you might want to have a talk with them. If not, you are the de facto maintainer
[19:52:42] <nanonyme> (this is why abandoned by upstream packages gets kicked out, packagers typically do not want to end up being maintainers)
[19:53:08] <paroneayea> nanonyme: well, kombu is maintained, but is relying on this out dated thing
[19:53:16] <paroneayea> nanonyme: anywy, yes, it's a problem
[19:53:43] <nanonyme> paroneayea, what was the library it depended on again, anyjson?
[19:54:30] <paroneayea> nanonyme: yes, anyjson
[19:54:39] <paroneayea> https://pypi.python.org/pypi/anyjson
[19:55:08] <nanonyme> I'll go poke some people
[19:55:41] <paroneayea> nanonyme: thanks :)
[19:55:46] <paroneayea> nanonyme: ping me if you get results!
[20:07:45] <nanonyme> paroneayea, well, kombu is part of Celery project, so tried to raise the matter with one of Celery lead devs. Maybe he has an opinion
[20:13:29] <paroneayea> nanonyme: cool