Pywcs - #13

Also -- the docstrings need to be converted to Numpydoc format.

This is now ready for merging. AFAIK, it meets all of the guidelines.

It does not work with Python 3.x (setup.py in master doesn't currently work, and I think that's better left fixing elsewhere).

I don't expect a full review -- I'm sure we'll find other problems over time.

However, of particular interest to @iguananaut and @eteq are the changes to the setup.py script. It walks the astropy tree looking for setup_package.py scripts and imports them. Then it looks for special functions within each that allow customizing C extensions, package_data and data_files. If you can think of a better way, let me know. It also checks for the existence of Numpy itself (rather than letting setuptools do it), since Numpy is needed during the astropy.wcs build -- if you don't check for it you get an ImportError on Numpy which isn't as helpful for the end user.

On the setup.py stuff: Looks fine to me, though it's already getting to the point where it would be worth moving parts of it into the setuputils module we discussed. I'd say go ahead and commit the changes now as is, and we can move stuff later (I'm thinking now maybe we should call it setup_helper.py for symmetry with version_helper.py).

Oh, I see you already created setuputils.py. That's fine then--just more things can go into it I think. And I'm still for renaming it to setup_helper (or helpers?).

I have not reviewed all the code, but I installed it and ran astropy.test() and all the tests run fine.

The api.rst file still mentions the old testing framework (including nose). To run specifically the wcs tests, the preferred way is now:

import astropy
astropy.test('wcs')

Otherwise it looks good to me.

Ok, I'm fine with this now! I'll let @eteq and/or @perrygreenfield review as well before we merge.

We may want to re-order how the documentation works, but that can be changed later when more of the documentation is in place. Also, when I run the docs, I see

/Users/erik/src/astropy/docs/wcs/references.rst

Should that worry me?

I also get:

 checking consistency... [snip]/astropy/docs/wcs/references.rst:: WARNING: document isn't included in any toctree

It looks like references.rst is included in many files, but just not in the toctree. I think we can ignore this since it makes sense it's not included?

The "references.rst" not included in the doctree is annoying, but I can't find a better solution. "references.rst" contains a number of external hyperlinks that are referenced throughout the wcs documentation that are too cumbersome to repeat everywhere (they would be particularly annoying in docstrings). But there's not real "content" in references.rst. We could put these in the "conf.py" rst_epilog variable, I suppose, but I'm not a hugh fan of the modularity of that.

Hmm... I don't think it's a good idea to let warnings that we still know are "ok" slip into the doc build, because then people will start ignoring legitimate warnings that are because they mis-typed something (and its easy for sphinx to go crazy with warnings if you don't keep on top of it!). I don't know if there's a way to supress that warning? Alternatively, there's the rst_epilog feature in conf.py - we could put everything in a global reference file and load it from conf.py ... but I can do that separately if we judge that a good idea.

Oh, and I also prefer avoiding setuputils.py as a name to prevent confusion with the package by that name. setup_helpers.py seems fine.

I figured out a solution for the warning. Just rename references.rst to references.txt and update the include statements appropriately.

I also agree with using the name setup_helpers.py instead of setuputils.py

Thanks for the solution (to rename references.rst to references.txt). I think that's a good one.

Also -- I will rename setuputils.py to setup_helpers.py.

Looks good to me now.

Same here.