[Numpy-discussion] Wanted: new release manager for 1.5 and above

David Cournapeau cournape at gmail.com
Sat Jan 16 09:17:38 EST 2010

On Sat, Jan 16, 2010 at 9:59 PM, Ralf Gommers
<ralf.gommers at googlemail.com> wrote:
> On Sat, Jan 16, 2010 at 5:19 PM, David Cournapeau <cournape at gmail.com>
> wrote:
>> On Sat, Jan 16, 2010 at 4:57 PM, Ralf Gommers
>> > From working on
>> > the docs and scikits.image I am familiar with most of NumPy/SciPy, but
>> > not
>> > with the C internals.
>> That's not a problem - I was not either when I started doing it. And
>> there are still a lot of areas I am not familiar with. There is no
>> need to know everything about numpy to do a good job.
> Good, then I'll start learning.
>> One thing you could start doing is trying to make the mac os x dmg
>> from the paver script, and familiarize yourself with virtualenv if you
>> don't know it (I use virtualenv to install numpy in a temporary
>> directory before building the doc - this guarantees that the doc
>> matches the exact same version of numpy as the one you are packaging).
>> I spent some time cleaning the paver script before the 1.4.0 release,
>> so it should hopefully be readable.
> The paver script is indeed very readable. I'm familiar with virtualenv, and
> just tried building a dmg. So here's my first question:
> You use virtualenv with the --no-site-packages option. This means a whole
> bunch of stuff already present on my machine has to be downloaded again when
> building the docs (sphinx, numpydoc, pygments, jinja, etc).

Yes, that's because I want to be sure to get the wanted version for
sphinx. It may not matter much anymore, but before, there were a lot
of instabilities between sphinx/numpydoc/matplotlib extensions

> You get Sphinx
> 0.6.4 when you need 1.0 (I think), and doc generation fails because MPL can
> not be found.

You need matplotlib to build numpy doc I think - at least it used to
be the case. Maybe it is not needed anymore. As I have matplotlib
installed on my machine anyway, this has never been an issue for me.


More information about the NumPy-Discussion mailing list