[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: bug in rint() function

On 2004-12-13 17:49:30 +0100, Michael Schmitz wrote:
> Maybe your bug report would have been more convincing if the test case
> results were appended, and did actually show both the bug, and your
> proposed fix performing better than generic rint.

Yes, you're right. I'll add that...

> I ran a quick test, and find that your code doesn't work on ix86
> (myrint reproducing the input float there).

because x86 has extended precision (and the wrong result is in fact
due to a bug in gcc, which is not a conforming ISO C implementation
when the processor has extended precision).

> A glibc maintainer testing this on his home machine might have given
> up fast ...

He should see that the code is powerpc-specific. Anyway I didn't get
any mail from anyone.

> Other than that: no idea. Did you ask on the libc ML?

No, I'm not aware of this mailing-list.

Vincent Lefèvre <vincent@vinc17.org> - Web: <http://www.vinc17.org/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
Work: CR INRIA - computer arithmetic / SPACES project at LORIA

Reply to: