[Numpy-discussion] rc2 for NumPy 1.4.1 and Scipy 0.7.2

[Ralf Gommers]

On Sun, Apr 11, 2010 at 11:08 PM, Sebastian Haase <seb.haase at gmail.com>wrote:

> Hi Ralf,
> congratulation to the new RC.
> Just for the record, where did you announce RC1 ? Only on numpy-dev ?
>

Since I'm new to this and wasn't sure it was all going to work right, I
announced rc1 only on scipy-dev (which in my mind is roughly a subset of
numpy-dev). This one I should probably also announce on scipy/MPL/pygame
lists to get more feedback. Will do that today, with the below explanation.


> On Sun, Apr 11, 2010 at 11:09 AM, Ralf Gommers
> <ralf.gommers at googlemail.com> wrote:
> <snip>
> > For SciPy there will be no 2.5 binaries -
> > because 0.7.x is built against NumPy 1.2
>
> Could you elaborate on this ? I thought the new numpy should build
> with the new scipy ... ?
>

This "new" scipy works with the new numpy, but for backwards compatibility
is built against older versions. For python 2.6 scipy is built against numpy
1.3, for 2.5 it should be built against 1.2. I'm not so sure about the
reason for that difference, but that's how it was done for 0.7.0 and 0.7.1.

Note that these releases are based on the 1.4.x and 0.7.x branches, which
are already quite old. Especially scipy, that branch was created in Jan
2009. Releases based on currents trunks will be numpy 2.0 and scipy 0.8,
which are the next target.

These releases are maintenance releases to recover from the binary
compatibility issues that plagued numpy 1.4.0.


> > it does not work on OS X 10.6, and
> > on Windows I see some incomprehensible build failures.
> >
> > Binaries, sources and release notes can be found at
> > http://www.filefactory.com/f/4452c50056df8bba/ (apologies for the flashy
> ads
> > again).
>
> Why not put these on the sourceforge page ?    (I don't know anything
> about the technical issues,  only a thought...)
>

Will do that the moment I get the password.

Cheers,
Ralf


>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20100412/cda6d6a4/attachment.html>