Fwd: Re: possible runtime class intialization error

Tue Dec 3 01:51:00 GMT 2002

Martin> You got this result 0/109 in gcj as well as IBM VM? I tried
Martin> only Sun's HotSpot...
Sorry, I ran with gij and got the same result. All that means is that
it is a compiler bug.
Interestingly, the bug doesn't occur with `gcj -C'. It only occurs
when generating object code. Compiling from .class files also gets
the correct result.
Could you submit a PR and upload your .gz file along with complete
instructions? I can't debug this immediately, but we need to track
it.
I looked at the code a bit. It looks like another bug in computing
constant expressions. This is an area we aren't very strong on right
now :-(
If you're up for debugging you could try writing a smaller example
(shouldn't be too hard) and then tracking through
fold_constant_for_init.
Tom