[LLVMdev] Rework of Vector/Scalar Classification

David Greene dag at cray.com
Wed Dec 9 16:40:10 PST 2009

On Wednesday 09 December 2009 18:33, Dan Gohman wrote:

> > But until that happens, there is no alternative to the patch I submitted.
> True, but there isn't anything using this information
> (specifically, isVector flags on operands) anyway, unless
> there's something we don't know about.

No, nothing currently.

> > It's a bit frustrating to be told to rework a patch to operate in a
> > different (and better) fashion only to have it rejected.
> I've not followed closely enough to know who told you to do it
> this way; if they think the new patch is good then that's fine
> with me. It looks like it probably won't break anything.

Evan asked me to rework it.  I'm glad he did, for I learned a few new TableGen 
tricks.  So the exercise wasn't a complete waste.  I've decided I'm going to
file the patch away in a little "to-do" folder and when I need it I can pull
it back out.  Until then, I don't see a reason to push it upstream if no one
else finds it especially useful at the moment.

It's more important I get other stuff merged anyway, so I don't think we need
to do anything more with this right now.


More information about the llvm-dev mailing list