gcj patch ping

[Tom Tromey]

>>>>> "Per" == Per Bothner <per@bothner.com> writes:

>> http://gcc.gnu.org/ml/gcc-patches/2002-05/msg00983.html
>> Somewhat hacky fix for PR 6520

Per> This doesn't seem right, not without knowing what are
Per> the actual semantics for what fold is *supposed* to do for
Per> constants.  It seems to me wrong for fold to be modifying
Per> existing nodes.

That makes sense.  I'll investigate further.

>> http://gcc.gnu.org/ml/gcc-patches/2002-04/msg00393.html
>> Assertion facility

Per> I'm not familiar with the assert facility, but I trust
Per> your judgement.

Ok.  Basically this is a straightforward implementation of the new
feature.  I'll check it in.

>> http://gcc.gnu.org/ml/gcc-patches/2001-12/msg02182.html
>> Minor optimization in bytecode generation

Per> It seems ok.  But I wonder why you need to test:
Per>    && reloc->label != block

I believe I added that to prevent an infinite loop if the `goto's
themselves form one.

Per> Also perhaps add to the comment at the top of the loop
Per> your rationale - i.e. "this can happen when generating
Per> a 'finally' clause".

Ok.

Tom