This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC 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: Regression with "[PATCH]: Fix libstdc++ wchar failure, ada bootstrap"


> Date: Wed, 16 Jul 2008 15:32:38 -0400
> From: "Daniel Berlin" <dberlin@dberlin.org>

> I'm not going to worry about missed opt problems until the wrong code
> bugs go away.
> It happens to have taken a little longer than expected to shake some
> of the bugs out.
> If it makes you unhappy to see some FAIL's temporarily, XFAIL them. I
> won't stop you.
> 
> I personally think it's a waste of time to worry about it right now,
> because it's not a case where we plan on having them FAIL for some
> significant length of time (IE till the next release or whatever)  or
> I would have done it.

The problem, which I thought was obvious, is that when I see a
new regression, I have no idea whether it's a "missed opt
problem" or a "wrong code bug" and have to dig further to check
that it wasn't some badness exposed specific to my port.  I
guess other people tracking the state of HEAD, without also
digging deep into the forest, react the same.

If these regressions had been xfailed, I wouldn't even see them.
If they had been mentioned in the gcc-patches@ message, I'd have
found them.

Like the gfortran.fortran-torture/compile/pr32663.f SEGV the
other day, I could easily have missed that one.  If it had just
been mentioned which new "missed opt problems" were seen at
regtest, it'd have helped.  It's not even like you have to worry
about fixing them, since apparently the consensus is to let this
pass, well, let's say temporarily.

brgds, H-P


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