[SciPy-dev] [scikits] openopt SVN instable for the moment
Tue Apr 8 15:32:14 CDT 2008
2008/4/8, dmitrey <email@example.com>:
> Matthieu Brucher wrote:
> > Hi,
> > The current repository head of openopt is unstable because we are in
> > the process of enhancing its behaviour:
> > - There will be a registry where everyone will be able to... register
> > its own openopt compatible wrapper of an optimizer. This way, the
> > burden of developing these wrappers will not only rely on dmitrey,
> > everyone will be able to write and test new ones for one's favorite
> The problem is I just mere don't see those dozens of people with willing
> to contribute sufficient solvers to OO. Some people answered they could
> do it in the sake of money, but it's inappropriate neither to me no to
I didn't say dozens of people. If one developer contributes one wrapper for
a solver, it is worth it.
> - The sys.path list will not be modified by openopt anymore (ticket
> > #50 on the scikits trac). This should make "python setup.py develop"
> > possible.
> I still think the path issue must be done in more far future.
This has been an issue for more than a year because of a design issue. This
is now solved in the lastest revision of the scikit. It didn't take me an
evening. Other people asked for it, so this was high on my todo list. This
doesn't change a thing for you, except from the fact that you have to use
namespace, which is a good thing.
> > We have to check every solver so that it makes the correct imports and
> > add it to the openopt registry, and then everything should be fine.
> I still don't think it's a good idea
Openopt should not mess with the sys.path the way it does. Now, openopt has
a classic way of importing subpckages, and it will not change a thing for
the user. Even less namespace conflicts can happen now, and this is a good
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Scipy-dev