Split out OMP constructs' SIMD clone supporting code (was: Splitting up gcc/omp-low.c?)

Thomas Schwinge thomas@codesourcery.com
Fri Apr 15 14:33:00 GMT 2016


Hi!

On Fri, 15 Apr 2016 14:15:42 +0200, Jakub Jelinek <jakub@redhat.com> wrote:
> On Fri, Apr 15, 2016 at 02:11:45PM +0200, Thomas Schwinge wrote:
> > On Fri, 15 Apr 2016 13:57:05 +0200, Jakub Jelinek <jakub@redhat.com> wrote:
> > > On Fri, Apr 15, 2016 at 01:53:14PM +0200, Thomas Schwinge wrote:
> > > > For all the other splitting patches that I have posted/proposed, the idea
> > > > then is to commit these onto both gcc-6-branch and trunk?
> > > 
> > > If we branch today, then yes, though the gcc-6-branch commits would need to
> > > wait until after 6.1 is released.
> > 
> > Uh.  I fear these patches will be a bit of a pain to maintain if there's
> > then on-going development in trunk, touching the gcc/omp-low.c file?  I
> > thought we had agreed to get these cleanup changes into trunk in time for
> > the first GCC 6 release, to exactly avoid this kind of pain (that is,
> > divergence of the gcc/omp-low.c files in trunk and gcc-6-branch)?  :-( I
> > have not been aware that you intend to create the gcc-6-branch today, and
> > neither have I been aware that once the branch's created, the other
> > cleanup changes would have to wait until after the GCC 6.1 release.
> > There is no way I can get all this done today, or on the weekend.
> 
> Then postpone the rest shortly before 6.1 is released.
> We shouldn't have big changes to trunk before 6.1 is released anyway,
> otherwise we won't be able to test on the trunk fixes intended also for 6.1.

Hmm.  What about the further patches that I already have in the queue for
submission, just waiting for final testing/polishing?  Are you asking me
to rescind these now, and then have to reproduce the work later on (when
exactly?), or will it still be fine to get these approved and committed
in the near future (next days)?  Obviously, I'll strongly prefer the
latter, as I already have done most of the work, and as that's what I
understand to be the modus operandi we had agreed on.


Grüße
 Thomas



More information about the Gcc-patches mailing list