This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: COND_EXPR lowering patch
- From: Diego Novillo <dnovillo at redhat dot com>
- To: Jeff Law <law at redhat dot com>
- Cc: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Tue, 26 Aug 2003 08:27:25 -0400
- Subject: Re: COND_EXPR lowering patch
- Organization: Red Hat Canada
- References: <200308260712.h7Q7CQtu000953@speedy.slc.redhat.com>
On Tue, 2003-08-26 at 03:12, law@redhat.com wrote:
> Though given that I want to see us incrementally improve things, I do not
> want to see a mega patch which does both changes at once.
>
> In fact, I'd really like to slow the patches down and get resolution
> on those which are already outstanding, starting with the LOOP_EXPR
> patch. I believe there is a single outstanding issue with it that
> is preventing it from being installed (the gcov failures). I'd really
> like to see that resolved before submission of COND_EXPR lowering.
>
Agreed. Let's get some closure on the outstanding patches.
One other (minor) thing about COND_EXPR lowering. I would like us to
lower COND_EXPRs in a post-gimplification pass. COND_EXPR may be a
useful construct for things like program understanding. We have no use
for it in the optimizers, but maybe we find a use for it in other
analysis passes that may be developed in the future.
I don't feel too strongly about it, myself. One could argue that those
analyses should work on GENERIC. And right now I don't see any other
use for non-lowered COND_EXPRs, so if people think we should lower them
right away, I won't mind.
Diego.