whitelisting additional matherr funcs for conform tests

[Mike Frysinger]

On 22 Jul 2015 19:44, Joseph Myers wrote:
> On Tue, 21 Jul 2015, Mike Frysinger wrote:
> > would it be possible to whitelist these two funcs for the ia64 port ?
> > not sure if that's easily possible with the code as-is, and whitelisting
> > it for all arches seems wrong as it'd allow other ports to screw up too.
> 
> I think you need to whitelist it for all architectures.

OK

> I'd like eventually to obsolete matherr support (turn matherr / 
> _LIB_VERSION into compat symbols, give new symbol versions to the relevant 
> libm functions that go to new wrappers that just set errno without 
> allowing for matherr), but I suspect the part of that involving new symbol 
> versions that don't use matherr would be hard with the ia64 implementation 
> (matherr handling could still be disabled for static links, which would 
> deal with this whitelist, however).

i'm totally onboard with obsoleting these symbols and pushing people to the
conventions that are in POSIX (e.g. fenv.h).
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20150723/ed860eb4/attachment.sig>