[pydata-outreach-staff] Laptop setup guide, and questions

[Asheesh Laroia]

Status so far:

* I pushed to 
https://github.com/paulproteus/PyData-Workshop-Sprint/wiki/Laptop%20setup

* I read all about compilers on Mac OS, and wrote instructions for getting 
a compiler there. I believe I've made the best recommendation. It'll 
require a ~100MB download, except OS X 10.5 it's a ~1GB download. A C 
compiler is required to build pandas' Cython extensions. If this step goes 
wrong at the event, it could cost the attendee an hour or more.

* Linux is tested all the way through. Another tester to verify would be 
great.

* Windows is tested all the way through. Another tester to verify would be 
great.

* For OS X, I only have a OS X 10.5 laptop here to test with. I'd 
appreciate if someone else can write up instructions for using Anaconda 
Python, since that seems simplest, given that different versions of OS X 
have different base versions of Python bundled with them. If no one 
volunteers in the next hour, I can write those up later today.

Other notes:

* vid, I don't seem to have push access to your repo at 
https://github.com/svaksha/PyData-Workshop-Sprint . Maybe I'm doing 
something wrong? I'm trying to push as 'paulproteus' and getting the 
same old message.

* I'm impressed with how, just like it was hard to get the official 
compiler on UNIX systems in the 1990s, Apple makes it very difficult to 
get the official compiler on their UNIX. It requires creating and 
maintaining an account on their site,and it actually took me a few tries 
before account creation worked properly. To save time for attendees, I've 
cached a copy of the OS X XCode command line tools in an account of mine 
on the web.

* Other pandas maintainers fixed the Cython requirement so it works on 
0.15 or later, rather than requiring 0.17.1, so Debian/Ubuntu users can 
'apt-get install cython' and get it to work now!

* EPD Free doesn't include Cython, so EPD isn't a super helpful starting 
point.

* I'm pretty un-thrilled about the idea of giving Windows users "Anaconda 
Python", even if they already have Python installed. If they have Python 
already, they'll have to be careful to use Anaconda Python throughout the 
day, and that is probably an error-prone, habit-fighting procedure. Mac 
users, well, there's too many Python versions distributed by Apple to 
suggest otherwise.

* Because of the bug in distutils, I switched the Windows compiler to the 
Microsoft one, which I've tested and adjusted the docs for. 
(Interestingly, Anaconda bundles mingw32 compilers and presumably works 
around the distutils bug.)

* When you run out of space when using github's 'gollum' wiki tool, it 
corrupts your local git repo.

* Numpy's website is down: http://numpy.scipy.org/ ("404 / There isn't a 
GitHub Page here" is what I see there.)

* I still have to write the final part about configuring a Github account, 
and doing 'git remote rm origin' etc. to point the git remote "origin" at 
your fork, and create an 'upstream' origin. Should be done and pushed 
shortly.

* The text editor configuration guide uses 4 spaces for "tab". That 
seems consistent with the Pandas recommended style.

-- Asheesh.