CVS Commit
Matthew Fluet
fluet@CS.Cornell.EDU
Thu, 27 Sep 2001 18:30:31 -0400 (EDT)
> > O.k. I've re-tar.gz'ed my entire x86-codegen directory (sans CVS and CM).
> >
> > http://www.cs.cornell.edu/People/fluet/MLton/x86-codegen.tgz
> >
> > I think that earlier I had bundled the one from the src.tgz that Stephen
> > posted earlier.
> >
> > What's in that file works with what I have here. That's all I'll claim
> > for now.
>
> There were quite a few differences between that and what I had, in x86-liveness,
> x86-simplify, x86-codegen, x86-entry-transfer, x86-jump-info. Anyways, I
> dropped your directory and I still see exactly the same error on all
> regressions, using an NJ-G0.
>
> I put up my snapshot at http://www.star-lab.com/sweeks/src.tgz
I'm compiling and looking at it now.
I did a diff between my sources and the src.tgz, and I think I know
what the problem is:
[fluet@lennon fluet]$ diff -r --exclude=CVS --exclude=CM mlton/mlton-cvs-main/src/lib/ mlton/mlton-bug/src/lib/
diff -r --exclude=CVS --exclude=CM
mlton/mlton-cvs-main/src/lib/mlton/basic/directed-graph.sml
mlton/mlton-bug/src/lib/mlton/basic/directed-graph.sml
1156c1156
< setForestNodeInfo (n', {loopNodes = [graphNode n],
---
> setForestNodeInfo (n', {loopNodes = [],
My sources are the top, yours are the bottom.
BTW, I also noticed that I left the call to shrinkExp commented out in the
fixed contify.fun.