[15:41:39] <dshap> Hey all - anyone know if I “pip install -e some_git_url” and it checks out the source into my /src directory in my virtualenv. and then i pip uninstall it, it should delete the directory it created within /src, right? mine is not
[15:42:02] <dshap> all it’s doing is deleting the egg-link file within site-packages
[17:36:34] <aoberoi_> okay does that mean its a best practice to use a separate virtualenv for module developers (not just applications)
[17:37:43] <aoberoi_> and i'm trying to set up travis-ci for my module too, so i'm not sure what the install and test scripts look like if thats the case
[17:38:01] <qwcode> many people use tox, so I guess they can get away with not needing a virtualenv for development, because tox handles all that per environment
[17:39:25] <aoberoi_> hmm just looked up tox, it looks like it handles the matricies of environments, i'm not sure how that would work on travis since it also owns that aspect
[17:40:59] <qwcode> people often maintain both, or pip for example combines them https://github.com/pypa/pip/blob/develop/.travis.yml
[17:41:58] <aoberoi_> oh wow, that seems kind of complex
[17:42:07] <qwcode> tests_require is for when using "setup.py test"
[17:44:40] <aoberoi_> okay, and right now thats what i'm using as the script for travis as well as what i run locally while developing
[17:45:14] <aoberoi_> actually another point of confusion is what nose and the nosetests CLI is actually meant for if setup.py test runs tests
[17:46:19] <elarson> aoberoi_: you can configure a setup.py to use nose, pytest, unittest, etc.
[17:47:07] <elarson> aoberoi_: if it helps, here is how I configure a module I wrote https://github.com/ionrock/cachecontrol/
[17:48:53] <Ivo> aoberoi_: I'd start off with travis' simpler examples - http://docs.travis-ci.com/user/languages/python/
[17:49:32] <Ivo> move to doing other things when you find you need to
[18:00:41] <qwcode> aoberoi, here's another breakdown of install_requires vs requirements http://packaging.python.org/en/latest/technical.html#install-requires-vs-requirements-files
[18:00:57] <Ivo> you might also have a dev-requirements.txt with extra stuff for testing / development that travis would want to install (or maybe not)
[18:01:21] <Ivo> if you told travis to use python setup.py test, then setuptools would install things in tests_require list
[18:01:46] <qwcode> btw, the link I just posted is fairly new, so if anyone wants to critique it, feel free, and post an issue to the PUG https://github.com/pypa/python-packaging-user-guide
[18:02:36] <Ivo> qwcode: I reckon it could be broadened to the various kinds of requirements one lists
[18:04:04] <aoberoi> great stuff guys! i'm going to try to ingest this and i'll follow up if i have more questions
[18:04:43] <aoberoi> one more thing while i have someone's attention
[18:05:25] <aoberoi> in 2.7 there are plenty of assertion methods provided by unittest.TestCase, but those don't exist in 2.6
[18:05:54] <aoberoi> i thought that was something nose would help me with but it seems like nose just takes whats available in unittest.TestCase and changes the naming convention to snake case
[18:06:35] <aoberoi> then i ran across this package called sure : https://pypi.python.org/pypi/sure/1.2.5
[18:07:25] <Ivo> well all the testing frameworks are bound to have one or two quirks of their own
[18:07:36] <aoberoi> what would be a recommended way to solve this? i feel like i keep finding nose less and less useful and i think thats surprising since it seems so ubiquitous in the community
[18:12:40] <aoberoi> i don't see how it helps write the tests... i feel like i'm missing something
[18:13:01] <aoberoi> and executing the tests can be done by just invoking `python setup.py test`, no?
[18:13:47] <Ivo> if you provide a test command class to integrate into whatever test suite you're using
[18:14:09] <Ivo> IIRC nose originally came about when python's in-library test library was a lot weaker than it currently is
[18:14:52] <Ivo> But I imagine pytest and nose and unittest2 can all get exactly the same things done in 80% of use cases if not more
[18:16:33] <aoberoi> maybe what i should do is to write my tests with just unittest.TestCase and when i actually run up against something i can't do, i'll see why those others are useful
[18:18:07] <Ivo> out of the box pytest and nose provide nicer test runners
[19:57:15] <aoberoi> general question about the article... the general idea is that if i know the concrete dependency, i should just pre-empt the resolution by specifying it before the line: '-e .', right?
[19:59:11] <dstufft> aoberoi: i'm not sure what you're asking and i'm about to step out to pick up my daughter
[19:59:17] <dstufft> I updated the blog post though
[19:59:34] <aoberoi> okay no worries, thanks for your help :)
[21:12:36] <aoberoi> is there a difference in how setuptools loads modules in python 2 and python 3?
[21:13:18] <aoberoi> concretely: https://travis-ci.org/aoberoi/Opentok-Python-SDK/builds/21538838, i've got just a couple trivial tests and it won't load in python 3 at all
[21:13:56] <aoberoi> corresponds to this code: https://github.com/aoberoi/Opentok-Python-SDK/tree/253f1949dafc7577bd3a4360c435e81e3d10fc35
[23:10:03] <Ivo> aoberoi: it worked on python3 when i renamed OpenTokSDK.py to sdk.py and changed the associated __init__.py
[23:10:56] <Ivo> aoberoi: you're also putting markdown in your README.rst
[23:14:22] <aoberoi> Ivo: haha yeah i just sloppily renamed, planned on coming back later to that
[23:15:00] <aoberoi> interesting, so does that mean python3 doesn't like capitalized module names?