This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: XFAIL tests that aren't regressions
On Wednesday 02 May 2001 02:44, Mark Mitchell wrote:
> >>>>> "Franz" == Franz Sirl <Franz.Sirl-kernel@lauterbach.com> writes:
>
> Franz> To be honest, I find the current situation with the branch
> Franz> a little bit frustrating. On powerpc-linux-gnu I've nearly
>
> Me, too. :-)
>
> Franz> more feedback would be nice, _especially_ when we are
>
> Do you mean more feedback from me? If so, I'm sorry -- I've been very
> busy with both the release and other projects. But, ask me anything,
> and I'll try to answer!
No, middle-end, not frontend. Besides the 2 c++ showstoppers (you fixed
static11 yesterday), I only have about 9 rh71/rawhide packages on my local
machine with c++ compile errors left, openjade and freetype probably the most
important ones (no further analysis done yet).
> Franz> Well, I just fixed the 73 FAILS for ObjC (if someone OK's
>
> Great!
>
> Franz> Another point on loop-2b, why not just disable this
> Franz> particular optimization on the branch? Seems more correct
> Franz> to me.
>
> This failure has always been there. We only want to fix regressions
> at this point. Perturbing things is *always* risky. Look at Nathan's
> innocent template/namespace-fixing patch this morning; what looked
> simple broke libgcj. We want to be as conservative as possible while
> still fixing the regressions.
Yeah, I know the risk, but from my analysis the failing point is a quite
small code hunk and it seemed easy to fix/disable to me, maybe I'm wrong.
Franz.