pmxbot IRC Log Viewer

[09:53:04] <aclark> dstufft: could vanity service be broken again? I'm not seeing hourly or even daily updates to the stats

[11:35:50] <nanonyme> Hmm, is there any way to define requirements only for a certain OS or a Python version?

[11:36:20] <nanonyme> The amount of requirements files I'm having is starting to blow out of proportion supporting different platforms

[11:38:05] <nanonyme> Having Python2 vs Python3 vs Linux vs Windows separate and keeping all in sync is pretty much maintenance hell

[11:46:21] <ionelmc> nanonyme: see http://pip.readthedocs.org/en/latest/reference/pip_install.html?highlight=marker#requirements-file-format

[11:46:35] <ionelmc> those are environment markers

[11:47:02] <ionelmc> nanonyme: see https://www.python.org/dev/peps/pep-0426/#environment-markers for what's supported

[11:57:50] <[Tritium]> oh sweet, no more chaining requirements files

[12:06:03] <mgedmin> it's a setuptools feature, right?

[12:06:17] <mgedmin> i.e. will it work with zc.buildout?

[12:10:14] <[Tritium]> it works in setup.py apparently

[13:35:54] <linovia> Hi

[13:36:21] <linovia> I'm a bit annoyed with some egg directory that is poping during a script execution

[13:37:11] <linovia> I get something in .python-eggs/SQLAlchemy-0.8.0-py27-linux-x86_64.egg-tmp in the current directory

[13:37:49] <linovia> Any hint on how to avoid this tmp egg creation ?

[13:39:55] <DanielHolth> which script

[13:40:54] <linovia> well, it's a full python application

[13:41:09] <linovia> what annoys me is that this egg-tmp is created during runtime

[13:41:47] <mgedmin> smells like a zipped egg install + pkg_resources.get_resource_filename()

[13:42:12] <mgedmin> my advice: don't use zipped eggs

[13:42:57] <mgedmin> which basically means preferably use pip, but if you must use easy_install/zc.buildout, use ... uhh, I forgot the option flag

[13:45:59] <linovia> mmm

[13:46:39] <linovia> It would be great if we could make setuptools use pip

[14:02:08] <ionelmc> linovia: setup.py install --single-version-externally-managed

[14:02:22] <ionelmc> or use pip to install

[14:02:36] <ionelmc> but isn't the .python-eggs some buildout thing?

[14:03:04] <linovia> ionelmc: well, this was built with dh_virtualenv which does a setup.py install

[14:03:31] <ionelmc> linovia: maybe the package has sqlalchemy in setup_requires

[14:04:09] <linovia> It does since I've been moving form the requirements to setup_requires so that we can pip install our integrated applications

[14:05:06] <linovia> as opposed to pip install -r requirement

[14:06:00] <ionelmc> linovia: why don't you use install_requires then?

[14:06:59] <linovia> ionelmc: my bad, it's in install_requires

[14:07:14] <linovia> we don't have setup_requires

[14:07:36] <linovia> I'm sorry I confused both

[14:07:50] <ionelmc> now i'm confused

[14:07:52] <ionelmc> :-)

[14:08:12] <linovia> the sqlalchemy is declared in the install_requires section

[14:08:50] <ionelmc> linovia: you could trace or patch up the code to see who exactly creates that

[14:08:55] <ionelmc> it's not that hard

[14:09:45] <linovia> another thing that I don't get is the SQLAlchemy is already installed in the virtual env as an egg

[14:10:07] <linovia> ionelmc: will try to trace that

[14:10:32] <ionelmc> linovia: take a look at https://pypi.python.org/pypi/hunter

[14:11:07] <ionelmc> another way is aspectlib but it's more tricky to setup (you'd need to do it in a usercustomize.py)

[14:11:56] <linovia> thanks for the pointers, going to dig this issue further

[17:16:13] <dstufft> aclark: it could be of course, though the disks aren't foull (what generally caused breakages in the past)

[18:38:38] <aclark> dstufft: weird, it typically updates hourly or so IIRC

[18:38:57] <dstufft> there's a cronjob that runs every hour yea

[18:39:01] <dstufft> is it still not updating?

[19:20:18] <buzzedword> hey folks! question-- i'm currently setting up an automated stack for my python devs. i've run into an issue where a) requirements.txt does not list requirements in dependency order, resulting in a hard crash on install

[19:20:55] <buzzedword> and b) on arbitrary apps, some of the packages are unverified, and i need to explicitly name the package to be verified

[19:22:00] <buzzedword> has there been any documented process to install all requirements in a requirements.txt and their dependencies?

[19:22:30] <buzzedword> also-- just as you can allow-all-external, can you do something like allow-all-unverified?

[20:06:39] <ronny> buzzedword: recent pip versions allow you to list --foo lines in requirements.txt

[20:07:01] <ronny> buzzedword: if dependency order is an issue the packages are broken and you need to provide a fix to them

[20:08:03] <buzzedword> ronny: by provide a fix, i'm assuming you mean patching upstream

[20:09:01] <ronny> yes

[20:09:18] <ronny> in the meantime using a devpi server and pushing local versions can elevate the problem

[20:10:37] <buzzedword> ronny: so, one of the most recent examples i can cite-- using a data science setup to install package "mrec" which depends on "scipy" -- scipy is also defined in requirements.txt

[20:10:50] <buzzedword> you're saying if dep order breaks here, the package itself is busted?

[20:11:15] <buzzedword> just trying to make sure i understand correctly

[20:11:38] <ronny> buzzedword: if it needs something for install/setup, it should have it in install_requires or setup_requires

[20:11:45] <buzzedword> ah!

[20:11:53] <buzzedword> ronny: okay, that clicks. thank you

[20:13:07] <buzzedword> so to your other point, if a package is untrusted, newer versions of pip can drop a --foo line to trust itself?

[20:13:25] <buzzedword> sorry-- s/untrusted/unverified

[20:13:49] <ronny> yes

[20:14:34] <ronny> also all packages not on pypi - please mail them regular for unsecure messy download fuckups, if enough people do that they might stop

[20:15:21] <buzzedword> sorry-- can you clarify a bit what you mean there?

[20:15:46] <buzzedword> as in, if a package is broken, mail pypi or the package owner?

[20:16:38] <apollo13> package owner

[20:17:30] <buzzedword> got it, that makes much more sense

[20:20:06] <buzzedword> thank you for the help!

[21:32:51] <Ashald> Hi

[21:33:18] <Ashald> I have a problem with creating a python package - is it the right place to ask a question?

[21:39:18] <Ashald> here is what I'm trying to do https://dpaste.de/H22v

[21:39:21] <Ashald> then I run this with python setup.py sdist https://dpaste.de/6d40

[21:39:21] <Ashald> (everything is OK)

[21:39:21] <Ashald> but when I'm running it for the 2nd time

[21:39:21] <Ashald> https://dpaste.de/0sn7

[21:39:22] <Ashald> This DistributionNotFound error occurs

[21:44:46] <Ashald> and when I'm trying to pip install it I have following error: https://dpaste.de/ZUkY

[21:44:52] <Ashald> What am I doing wrong?

[21:45:25] <Ashald> Or I cannot use entry_point with install_requires?

[21:47:01] <Ashald> is there is a way to fix it without adding six to setup_requires?

[22:21:22] <ionelmc> Ashald: do you actually need the file_finders thing?

[22:50:49] <Ashald> that's the whole purpose of the package

[23:38:18] <aclark> dstufft: working now \o/

Log file Viewer

Help | Karma | Search:

#pypa logs for Thursday the 2nd of April, 2015