[SciPy-dev] updating the numpy/scipy versions in Linux distros

David Cournapeau david@ar.media.kyoto-u.ac...
Wed Feb 18 00:55:50 CST 2009

Joe Harrington wrote:
> For the second part of the problem, of course there's no reason for us
> to document what Ubuntu or any other distro does to cut one of their
> releases.  What's needed is to document how our stuff gets into their
> release.

I think there is little to document, because there is no formal process
between "us" and "them". There is a lot of process within a
distribution, what's go where and when for which version, but that's
documented on each distribution.

>   If they pull, who does it, and what's the best way to signal
> them that it's time to pull, given that they're not on our lists and
> that for many distros, the person responsible for the package might
> not use or even personally care about our software.

In my experience, what matters the most is maintainer time (I sound like
a broken record, don't I ? :) ). If they have some time, they will
update - but there is the problem that the window to update a version
does not always fit with maintainer 'free' time. Then, there is the
incentive of users asking for it.

One thing to keep in mind, which may not be obvious to everyone: our
goal, as numpy/scipy developers, and the distributions goals are not the
same, if not antithetic. We care about distributing the most recent
version, and they care about stability and the least work possible. Not
updating is almost always easier than updating for them. For example,
some RH developers are really pissed about python 3k breaking a lot of
stuff, and would even want to see python 3k failing so that they don't
have to deal with the numerous maintenance problems (see
http://lwn.net/Articles/310450/). Another example is that debian
developers would like to split numpy in many small pieces, because numpy
is too big; from our POV, that does not make any sense.

My own opinion is that the only solution is to have our own packages,
published when we can; distributions then do what they want/can on their
own schedule.

> This is actually good news.  Maybe we'd be more up-to-date in the
> distros if we had some volunteers act as liasons between the Packaging
> Team and the distros.  All they'd have to do would be to follow the
> instructions on the Distros page to ping the distros whenever we cut a
> new release.  There are plenty of newish folks who have asked how they
> could help, but who can't actually contribute code or docs yet.  This
> would be a good task for them.  It would only take one or two people.

I am not sure packaging counts as a good task for newcomers. Packaging
is difficult to do well, there are a lof of small details to keep right,
and there is a lot of politics involved, which does not sound as the
kind of things newcomers would like to do.



More information about the Scipy-dev mailing list