Re: [patch,complex] Replace hard-coded constants

[Ralf Corsepius <ralf dot corsepius at rtems dot org>]

On 11/29/2010 06:10 PM, Eric Blake wrote:

> If you have concrete examples of something that should be in POSIX but
> is currently missing,

Short summary of this thread's issue:

OpenGroup's
http://www.opengroup.org/onlinepubs/9699919799/functions/catan.html
http://www.opengroup.org/onlinepubs/9699919799/functions/ctan.html
are vague wrt. the "RETURN VALUE" and error-handling of these functions.

They leave the implementors of these functions unclear about how to 
treat mathematical poles and numerical limitations of these function's 
implementations, but defer implementors to "ISO-C".

This is different when comparing OpenGroups's specification of these 
function's against their non-complex counter-parts:
http://www.opengroup.org/onlinepubs/9699919799/functions/atan.html
http://www.opengroup.org/onlinepubs/9699919799/functions/tan.html

> it would be a good idea to raise that as a defect
> in the POSIX standard to get that situation corrected.  Unfortunately, I
> haven't been following this thread closely enough myself to take on the
> task.

Ralf