[libcxx][PATCH] remove implementation of fmaf - fix for PR18910

Yaron Keren yaron.keren at gmail.com
Wed Mar 5 08:34:32 PST 2014


Hi,

Systems without fmaf will very likely miss fmal and all the C99 functions,
so they have many other problems to worry about.

LGTM.

One question, though. Why

 using ::fma;
 using ::fmaf;

but not

 using ::fmal;

This is not fma specific question - cmath repeats this pattern for all C99
functions.
Shouldn't all three version treated the same?

Yaron







2014-02-22 17:44 GMT+02:00 Marshall Clow <mclow.lists at gmail.com>:

> http://llvm.org/bugs/show_bug.cgi?id=18910
>
> Remove the implementation of "std:::fmaf" from libc++'s <cmath> header,
> and use the c library's one instead.
>
> I believe this bug is a historical relic from the early days of libc++
> implementation.
> It was part of the original libc++ checkin to the LLVM repo.
> Probably very old versions of Apple's C library were based on C89, which
> didn't have fmaf, and so Howard just provided this here.
> [ When I say very old, I mean in the Mac OS X 10.3/10.4 timeframe ]
>
> Now we have C99 libraries in most places, and they all (as far as I can
> tell) provide fmaf themselves, and (I think) we can remove this.
>
> If you're running on a system that has an older C library implementation,
> this could be a breaking change, and I want to know about it.
> I can put this code back under an #ifdef for those systems, but not if I
> don't know which systems they are.
>
> -- Marshall
>
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140305/58250064/attachment.html>


More information about the cfe-commits mailing list