[PYTHON MATRIX-SIG] Release 0.34 now available
James Hugunin
jjh@Goldilocks.LCS.MIT.EDU
Mon, 12 Feb 96 15:08:07 EST
The major news for this release is that Chris Chase's grammar patches
to implement multidimensional slices are included. This means that
you can write:
a[1:, :-1] and other exciting variations now work. The old keywords
All, Reverse, and Slice no longer work in this new version.
I should warn people that I rushed out this version because a few
people were particularly eager to get there hands on it. I haven't
tested this as well as I'd like to, so as usual use at your own risk.
I hope to release something more stable by this Friday, so if you're
not into the cutting edge...
It's reasonably likely that the implementation details of
multi-dimensional slices will change before they become an official
patch to the python core. Nevertheless, it's very likely that the
syntax/semantics of these slices will be exactly as they are in this
distribution.
Personally I find that my own code looks a lot cleaner without all the
Slice(None, -1) stuff in it, thanks Chris!
As usual, this release also includes many minor bug fixes.
(It also includes a small performance optimization, but its unlikely
you'll notice that).
Also, as of this release I officially give in to the will of the sig
on the type coercion issue. Paul DuBois's comment on the relative
performance differences between floats and doubles was probably the
straw that broke my camels back.
floats are now automatically coerced to doubles, and all the other
coercions that you'd expect.
I really like Konrad's idea of adding some hooks into the debugger to
detect automatic type coercions and I'm looking into that.
Also, this release should compile on the Mac (thanks to Steve
Spicklemire for making this happen). He's also got a version working
under windows, but I didn't have the time to make the necessary
changes before this release. Now if I can just get somebody to test
this on a BeBox!
Patches from 0.33 are in
ftp://sls-ftp.lcs.mit.edu/pub/jjh/NumPy.patch-0.33-0.34.gz
The whole thing is in
ftp://sls-ftp.lcs.mit.edu/pub/jjh/NumericPython-0.34.tar.gz
Enjoy - Jim
=================
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
=================