[12:09:01] <dropdrive> On 2.6, am I right in thinking that if `a.b.c` is a namespace package (i.e. a/__init__ and a/b/__init__ is the boilerplate call to extend_path), it's okay to have another package `a.b` that has real code in a/b/__init__? I think this is because of the existence check for __init__.py in nspkg.pth, yes?
[17:27:17] <bebop-haiku> hello everyone. I have been updating my vagrant/puppet files to use ubuntu 14.04 and python 3.4 and noticed that using pip -e <virtualenv> no longer works. I am having some trouble trying to get pip to install into a virtualenv, as I cannot source the activate script. Does anyone have any advice on how to make pip work in a similar way to the -e option?
[17:27:58] <apollo13> bebop-haiku: can you provide more information than "no longer works"?
[17:28:12] <bebop-haiku> It no longer appears to be a command line option
[17:28:39] <bebop-haiku> pip3 -e virtualenvs/kala gives the error no such option: -e
[17:28:57] <apollo13> ah same here for pip2 it seems
[17:29:33] <bebop-haiku> hum, it used to work just fine to specify a virtualenv
[17:30:11] <apollo13> does pip install --root work?
[17:31:06] <bebop-haiku> I tried that but it pip said that the requiements were already satisfied in /usr/local/lib/python3.4/dist-packages, I guess I could try to uninstall all those packages and see if --root works
[17:40:23] <bebop-haiku> dstufft: thanks that explains it
[17:40:35] <bebop-haiku> I was using pip from 12.04 lts
[17:41:02] <dstufft> bebop-haiku: That was version 1.0, 14.04 has 1.5.4
[17:41:21] <dstufft> https://github.com/pypa/pip/blob/develop/CHANGES.txt you might want to read over this to see what has changed, especially the ones marked **BACKWARDS INCOMPATIBLE**
[17:41:36] <dstufft> ignore the 1.6.0.dev1 entry though
[17:41:39] <dstufft> that hasn't been released yet
[17:41:58] <dstufft> Probably a better link https://github.com/pypa/pip/blob/master/CHANGES.txt
[19:08:45] <winsoff> Why are recursive acronyms so funny?
[19:16:05] <wsanchez> Can I ask pip not to write to ~/.pip?
[19:16:56] <dstufft> wsanchez: do you want it to write somewhere else, or just not write at all
[19:18:06] <dstufft> I think the answer to the first one is probably --log-file=some/other/file/path, I think the answer to the second one is Nope, please open a ticke
[19:20:59] <wsanchez> Our build system runs as root (Whoo!) sets up a sandbox to prevent writing to stuff outside of designated locations.
[19:33:11] <dstufft> jezdez: yes I think there was, I think qwcode was working on it, not sure what came of it
[19:35:21] <qwcode> jezdez, dstufft, I ended up at least documenting them (http://www.pip-installer.org/en/latest/reference/pip.html#logging) and surfacing the one option that was hidden
[19:35:59] <jezdez> we could rename --log-file to --error-log
[19:39:12] <dstufft> TBH I'm not sure we even need either of them, It seems like an appending log is easier implemented as pip install foo >> /path/to/logfile or pip instal foo | tee -a /path/to/logfile and the "error log" is better implemented as pip install -vvv foo > /path/to/logfile or pip install -vvv foo | tee /path/to/logfile
[19:39:55] <dstufft> (I get what the idea of the error log is, but I'm not sure anyone ever actually looks at ~/pip/pip.log to see the debug view of the last command rather than just run it again with -vvv)
[19:41:08] <dstufft> (also I like the idea of having less code and less options :D)
[19:43:47] <qwcode> yea, I admit I punted due to not wanting to stress over breakages and fallout with build systems using the current log options
[20:26:43] <wsanchez> Looks like pip invokes a python -c thingo and *that* wants to write to ~/.pip: python -c "import setuptools, tokenize;__file__='/...../CalendarServer/build/python-ldap/setup.py';exec(......)" install ......
[20:27:05] <wsanchez> I'm sorry for whoever had to write that.
[20:33:15] <qwcode> wsanchez, pip writes to it's log files using pip.log