Re: patch ping for altivec.h and vmx tests

[Janis Johnson <janis187 at us dot ibm dot com>]

On Thu, Jul 22, 2004 at 03:47:38PM -0700, Ziemowit Laski wrote:
> 
> On 22 Jul, 2004, at 15.21, Aldy Hernandez wrote:
> 
> >On Thu, Jul 22, 2004 at 01:26:16PM -0700, Janis Johnson wrote:
> >>Ping for several altivec patches.
> >>
> >>changes to altivec.h:
> >
> >Zem, could you be more careful next time?  Most of these correct 
> >problems
> >with the compile time error code, and the bool code.  At least I think
> >it was you.
> 
> Possibly; I can't quite remember. Of course, the reason it could have
> stayed broken for so long is because there were no appriopriate test 
> cases
> exercising this functionality, and so it boggles the mind why you would
> approve these four patches without additional test cases.

The new functionality is covered by checks in gcc.dg/vmx/ops.c, which
failed for lots of reasons.  One of the testsuite patches removes bogus
pieces of that test.  I should have mentioned that I also tested the
altivec.h changes using a C++ version of ops.c; I'll look into adding
that to the C++ testsuite.

Janis