An apology [Was: Floating point computation "non-bug"]

[Dave Elcock]

Dave Elcock wrote:
 >
> This is a serious and very frustrating flaw in an otherwise superb 
> compiler suite. Why does the problem keep getting brushed aside like 
> it's nothing to do with GCC?
> 

Everyone on the GCC team, please accept my humble apologies. There are 
no grounds for this accusation. The only serious and frustrating flaw is 
my own ignorance. I imagine many of you following this thread were 
either smiling knowingly and shaking your head, or rolling your eyes 
back in contempt. I think I jumped in at the wrong end of the pool, and 
it's a lot deeper than I thought, even at the shallow end.

Anyone out there about to make the same mistake should read the 
"Problems with floating point computations" paragraph within 
http://gcc.gnu.org/bugs.html under the section "Non-bugs". It's there 
for a reason. You really ought to take the time to read the linked paper 
about FP limitations, "What Every Computer Scientist Should Know About 
Floating-Point Arithmetic". It may seem a bit heavy in places, but it's 
essential reading and might save you a little embarassment.

Cheers,
Dave Elcock.

"Mmm! Humble pie! My favourite."