[Python-Dev] Silly little benchmark

Guido van Rossum guido@digicool.com
Fri, 13 Jul 2001 11:33:54 -0400

> Tim Peters <tim.one@home.com>:
> > It's not BINARY_ADD, it's the PyNumber_Add() called by BINARY_ADD, which,
> > given two strings, calls binary_op1, which does a few failing tests, then
> > calls PyNumber_CoerceEx, which fails quickly enough to coerce, and then
> > pokes around a little looking for number methods, and finally says "hmm!
> > maybe it's a sequence?".
> This seems to contradict what Guido just said about
> centralised coercion having been removed. Is one or
> the other of us talking nonsense, or do we misunderstand
> each other?

It's complicated.  I didn't know everything that was going on when I
wrote that before.  Now I've seen a bit more.  PyNumber_CoerceEx() is
called in order to accommodate old-style numbers for backwards
compatibility (and for complex, which hasn't been converted to
new-style yet).

We could add a new-style numeric add operation to strings so that
s1+s2 takes an earlier path in binary_op1().

I also note that binary_op1() tries PyNumber_CoerceEx() even when both
arguments have a NULL tp_as_number pointer -- at the cost of extra
tests the call to PyNumber_CoerceEx() could be avoided.  (I guess
binary_op1() could add such a test at the top and save itself some

--Guido van Rossum (home page: http://www.python.org/~guido/)