Re: [tree-ssa vs lno] who is right?

[Dale Johannesen <dalej at apple dot com>]

On Mar 26, 2004, at 8:06 AM, law@redhat.com wrote:
> In message <1080315577.12834.102.camel@p4>, Andrew MacLeod writes:
> > On Fri, 2004-03-26 at 10:30, law@redhat.com wrote:
> > > In message <1080308082.12528.27.camel@p4>, Andrew MacLeod writes:
> > > But again, I think step #1 is to figure out how we got them in the 
> > > first
> > > place.
> > > jeff
> >
> > It would be pretty easy, something like:
> >
> > maxmin_Result_140 = PHI <1(10)>;
> > T.55_142 = PHI <2(10)>;
> > <...>
> > maxmin_Result_133 = T.55_142
> >
> > could result in the 2 PHI nodes since we'd rename T.55_142 to
> > maxmin_Result_133 in hopes of getting rid of the copy.
> >
> > It could also happen if T.55_142 is later used as a PHI argument to a
> > maxmin_Result PHI result.
> Ah yes.  Seems quite reasonable.

Well, at least this wasn't a trivial question...thanks to all of you.

So the consensus is that copyrename's behavior is valid and reasonable,
and the only thing wrong here is the LNO branch's algorithm that assumes
otherwise.  Right?   I'll go fix that.   (The testcase is from sixtrack 
in SPEC;
it's small, though, I may be able to come up with something not 
copyrighted.)