[SciPy-Dev] Building docs in scipy?

[Pauli Virtanen]

Fri, 04 Jun 2010 19:30:57 +0800, Ralf Gommers wrote:
[clip]
>> Matthew helpfully pointed out some options (thanks!) to do this in git:
> http://news.gmane.org/gmane.comp.python.scientific.devel
> 
> However, none of the options he gave are automatic, so people will keep
> running into this. I think building docs should work out of the box, so
> I see 2 options:
>
> 1. we copy sphinxext in scipy. it's not like it changes often, so
> keeping things in sync manually is doable.

I'd say this is bad practice, and we should not do this.

Better to have a Makefile rule that checks it out from git before 
building the documents, if it comes to that.

> 2. sphinxext becomes a numpy module that can be imported and used from
> scipy.
> 
> Including it in the tarballs while it's not in the repo is not a good
> idea imho - tarballs should be exactly the same as the svn tag.

Matplotlib moved their sphinx stuff into a submodule. I'm a bit leery 
doing that either, since the Sphinx stuff has not much to do with Numpy 
itself...

-- 
Pauli Virtanen