internal compiler error in gcc-2.95.2/PPC with -funroll-all-loops

[Steven G. Johnson]

The infinite-loop bug in gcc on the PowerPC that I was encountering
( http://gcc.gnu.org/ml/gcc-bugs/2000-04/msg00829.html ) seems to have
disappeared in the latest CVS snapshot.

Or at least, I can't reproduce it any more using the same test case.  
Since no one specifically said that they fixed it, the cynical view would
be that the underlying bug is still there but is just no longer triggered
in this particular case.

For now, though, I'll assume that it's truly gone.  Thanks to whoever
(inadvertently?) fixed it!

Cordially,
Steven G. Johnson