Factoring algorithms on rtl ?

[Gábor Lóki]

Daniel Berlin wrote:
>> I have taken no position on this work.  I simply pointed out that it
>> appeared to be a generalization of cross-jumping, which is something you
>> didn't mention, so I wasn't sure if you were aware of it.
>>
> In this spirit, are you aware that df.c provides the same def-use 
> information you attempt to generate on your own?

Thanks for the info.
As I mentioned earlier, this patch is only in a very experimental phase.
The next version will use df.c if it's possible.