This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC] Add math tests for libieee


On Mon, 13 Feb 2017, Tulio Magno Quites Machado Filho wrote:

> Joseph Myers <joseph@codesourcery.com> writes:
> 
> > On Wed, 8 Feb 2017, Gabriel F. T. Gomes wrote:
> >
> >> The code that this patch changes is not executed, because the wrappers
> >> (in math/w_j0_compat.c and math/w_j1_compat.c) call __kernel_standard
> >> and return.
> >
> > It's executed with -lieee (meaning this is user-visible and should have a 
> > bug filed in Bugzilla accordingly).
> 
> In this case, should we be testing libieee too?

Since -lieee is an obsolescent feature I'd be wary of adding such tests.

In a world where _LIB_VERSION and matherr are compat symbols and libieee 
is removed, but without new symbol versions for existing libm functions in 
existing configurations of glibc, there is the issue of how to find issues 
hidden by the existing wrappers when testing on architectures that still 
use them in shared libm.  But the obvious approach for that would be 
testing static libm which in such a context would use the new wrappers for 
all types rather than the old ones.  (Right now testing static libm 
wouldn't achieve much, it's only in a case where static libm and shared 
libm use different wrappers for the same type that it becomes relevant.)

-- 
Joseph S. Myers
joseph@codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]