On Thu, Jan 15, 2009 at 11:53 PM, David Cournapeau <[email protected]>wrote:

> On Fri, Jan 16, 2009 at 12:18 AM, Darren Dale <[email protected]> wrote:
> > Hi Jarrod,
> >
> > On Wed, Jan 14, 2009 at 2:21 AM, Jarrod Millman <[email protected]>
> > wrote:
> >>
> >> Due to the fact that I was tired of adding site.cfg to scipy and numpy
> >> when building on Fedora and Ubuntu systems as well as a scipy ticket
> >> (http://scipy.org/scipy/numpy/ticket/985), I decided to try and add
> >> default system paths to numpy.distutils.  You can find out more
> >> details on this ticket:
> >>  http://projects.scipy.org/scipy/scipy/ticket/817
> >>
> >> I would like to have as many people test this as possible.  So I would
> >> like everyone, who has had to build numpy/scipy with a site.cfg
> >> despite having installed all the dependencies in the system default
> >> locations, to test this.  If you would please update to the numpy
> >> trunk and build numpy and scipy without your old site.cfg.  Regardless
> >> of whether it works or not, I would appreciate it if you could let me
> >> know.
> >
> > I tested on Gentoo, which does some non-standard stuff in an attempt to
> be
> > able to switch between implementations. The libraries in /usr/lib are
> > symlinks to the implementations in /usr/lib/blas/, and you can have
> various
> > implementations like atlas, reference, or threaded-atlas, so I can run a
> > program called eselect that lets me switch between the implementations.
> For
> > some reason, gentoo renames the f77blas libs to simply blas.
>
> That really baffles me. Why do distributions think it is ok to
> randomly rename libraries - they certainly could handle a eselect
> system without renaming libraries, so that they avoid breaking the
> softwares depending on the libraries,
>

I am in complete agreement. I've raised the issue with them before and didnt
make any headway. Maybe its time to try again.

Darren
_______________________________________________
Numpy-discussion mailing list
[email protected]
http://projects.scipy.org/mailman/listinfo/numpy-discussion

Reply via email to