On Jan 27, 11:42 am, Paul Rubin <http://[EMAIL PROTECTED]> wrote: > Arnaud Delobelle <[EMAIL PROTECTED]> writes: > > def simple_posmax(l, key=None): > > if key: > > return l.index(max(l, key=key)) > > else: > > return l.index(max(l)) > > def simple_posmax(l, **kw): > return l.index(max(l, **kw))
Yes. I wanted to make simple_posmax and clever_posmax as identical as possible to make the comparison fair. > > > simple_posmax is more than 3x faster on my machine. It's not > > surprising as even though the list is walked twice, it is all done in > > C and no new objects have to be created. Then only non-C bit is when > > the result of max(l) is fed to l.index(). > > It does expose a slight danger in that the list might somehow > self-mutate during the first traversal. I suppose. Although you'd have to be mad to have comparisons mutate stuff... Also, it only works with sequences. -- Arnaud -- http://mail.python.org/mailman/listinfo/python-list
