Use of C99 extra long double math functions after r236148

Stephen Montgomery-Smith stephen at
Tue May 28 12:03:20 UTC 2013

On 05/28/2013 06:51 AM, Stephen Montgomery-Smith wrote:
> On 05/27/2013 11:32 PM, David Schultz wrote:
>> Hi Stephen,
>> I wrote some tests to cover the corner cases for the complex
>> inverse trig functions. They don't find any nontrivial bugs in
>> your implementations. :-) Now that you have a commit bit, would
>> you like to commit your code, or shall I?
> I think I only have a commit bit for ports, not src.
> In any case, I would much prefer that you commit it.  I have a lot on my
> plate right now.
> Thank you for doing this.  It would be great to see this in FreeBSD.

Also, if I can brag a little, I think the only other implementation of
the complex arc-trig functions that is as accurate are the most recent
boost library implementations, and then only because I submitted bug
fixes to them.

I also found a bug in the Hull, Fairgrieve, and Tang algorithm for
cacos/cacaosh, which was faulty in certain extreme cases.  This bug is
documented here:

More information about the freebsd-numerics mailing list