fix pr23546, exposed by using VIEW_CONVERT_EXPR for vectors

[Laurent GUERBY]

20548 seems to be fixed on x86_64 (I've put it in WAITING for powerpc
and hppa where it was also reported), but two new regressions appeared,
those are probably unrelated to your patch:

23564: c52104f c52104h (run) missing check
23565: c32001e (run x86_64 only) inccorect array bounds

I didn't check 4.0 yet, will do within a few days.

Laurent

On Thu, 2005-08-25 at 09:35 +0200, Paolo Bonzini wrote:
> > For some reason I had this patch in my patch archive (well close, just
> > adding INTEGER_CST and not REAL), but I didn't write down from where it
> > came from. I'm pretty sure it's safe for Ada (it was in my tree for a
> > while), might fix a few things. I'll run the testsuite just after you
> > commit.
> 
> Committed to mainline and 4.0.
> 
> Paolo
>