Re: [tree-ssa] Java byte compiler status

[Andrew Haley <aph at redhat dot com>]

Jeff Sturm writes:
 > On Tue, 22 Jul 2003, Andrew Haley wrote:
 > > I've got functions as trees mostly working for the bytecode compiler.
 > > Unfortunately, I have a few test failures.  As far as I can see the
 > > front end isn't doing anything wrong, and when I disable optimizations
 > > on trees these tests pass.  Perhaps I'm generating trees that are
 > > illegal, but I don't think so.  I wonder if this might be some latent
 > > bug in one of the tree optimizers.
 > 
 > With an up-to-date tree-ssa build and your patch, I get:
 > 
 >         === libjava Summary ===
 > 
 > # of expected passes        3033
 > # of unexpected failures    15
 > # of expected failures      16
 > # of untested testcases     19
 > 
 > 7 execution failures, 8 output failures, all in bytecode->native
 > compilation.

Apart from one, these all seem to be happen with the tree optimizers;
if you disable tree-ssa the bugs go away.  It could be that I'm
generating bad trees, or perhaps the unusual structure of the trees
triggers some latent bugs in the tree optimizers.

Andrew.