[Numpy-discussion] Status Report for NumPy 1.1.0
Wed May 7 19:17:45 CDT 2008
2008/5/7 Anne Archibald <firstname.lastname@example.org>:
> 2008/5/7 Charles R Harris <email@example.com>:
> > On Wed, May 7, 2008 at 11:44 AM, Anne Archibald <firstname.lastname@example.org>
> > wrote:
> > > 2008/5/7 Jarrod Millman <email@example.com>:
> > >
> > > > I have just created the 1.1.x branch:
> > > > http://projects.scipy.org/scipy/numpy/changeset/5134
> > > > In about 24 hours I will tag the 1.1.0 release from the branch. At
> > > > this point only critical bug fixes should be applied to the branch.
> > > > The trunk is now open for 1.2 development.
> > >
> > > I committed Travis' matrix indexing patch (plus a basic test) to
> > > 1.1.x. Hope that's okay. (All tests pass.)
> > >
> > You should put it in the 1.2 branch also, then. I think fancy indexing and
> > tolist, which Travis mentioned as having matrix workarounds, need to be
> > checked to see if they still work correctly for matrices with the patch.
> Ah. Good point. I did find a bug - x[:,0] doesn't do what you'd
> expect. Best not release without either backing out my change. I'm
> still trying to track down what's up.
Okay, more tests committed. Travis has removed the .tolist()
workaround (which, it turns out, doesn't actually depend on
__getitem__). Not sure what needs to happen for fancy indexing -
there's some alarming jiggery-pokery involving self._getitem that I
don't understand. But these are not bugfixes anyway.
More information about the Numpy-discussion