From: Vincent St-Amour (stamourv at ccs.neu.edu) Date: Mon Dec 13 12:11:00 EST 2010 |
|
At Mon, 13 Dec 2010 17:29:11 +0100, Jos Koot wrote: > When multiplying any number (nan and inf excepted) by exact 0, should, of > course produce exact 0. If you're multiplying an integer that may or may not be zero (say, an index) by a float, you cannot assume that the result is a float, even though in all cases but one, it will be. If you're using unsafe operations for performance reasons, this can break if you're not entirely careful. Vincent
Posted on the dev mailing list. |
|