Tim Chase <[EMAIL PROTECTED]> wrote:
...
> >>> all = [(x==255 or y==255) and (255, 255) or (x,y) for (x,y)
> in itertools.izip(a1,a2)]
> >>> b1 = [x[0] for x in all]
> >>> b2 = [x[1] for x in all]
> >>> a1, a2 = b1, b2 # if you want them to replace the originals
>
> Seems to do what I understand that you're describing using "no"
> loops (other than those implied by list comprehension).
Yep, but the performance cost of the for-loops in the comprehension is
essentially the same as for such loops written "normally".
> There may be some nice pythonic way to "unzip" a list of tuples
> created by zip() but I couldn't scare up such a method, and
Perhaps what you have in mind is:
>>> a=zip('feep','grol')
>>> a
[('f', 'g'), ('e', 'r'), ('e', 'o'), ('p', 'l')]
>>> zip(*a)
[('f', 'e', 'e', 'p'), ('g', 'r', 'o', 'l')]
But this wouldn't help the OP all that much with his performance
problems with large 2-D arrays (though it required some guessing to
gauge that it _was_ Numeric arrays that he was dealing with;-),
Alex
--
http://mail.python.org/mailman/listinfo/python-list
