This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Java definite assignment
Alexandre Petit-Bianco <apbianco@cygnus.com> writes:
> It reports an error. Try building kawa-1.6.1 from sources for example.
That's sort of what I'm doing, except I'm working on my checked-out code.
(I'm hoping to get to the stage where I can use gdb to track down an
infinite loop in kawa.)
> > I assume my patch does not change this, since the change was only in
> > LOOP_EXPR, and there should be no LOOP_EXPR for your sample code.
>
> No, it doesn't. But it's in neighboring code which also happens to use
> `done_alternative'. I thought you could easily identify a similar
> quick and easy fix.
Today is your lucky day ... Could you test the following?
* check-init.c (check_init): Fix TRY_FINALLY_EXPR logic.
Index: check-init.c
===================================================================
RCS file: /cvs/gcc/egcs/gcc/java/check-init.c,v
retrieving revision 1.26
diff -u -r1.26 check-init.c
--- check-init.c 2001/02/04 22:44:02 1.26
+++ check-init.c 2001/02/06 21:24:45
@@ -571,8 +570,9 @@
{
words tmp = ALLOC_WORDS (num_current_words);
COPY (tmp, before);
- check_init (TREE_OPERAND (exp, 0), tmp);
- check_init (TREE_OPERAND (exp, 1), before);
+ check_init (TREE_OPERAND (exp, 0), before);
+ check_init (TREE_OPERAND (exp, 1), tmp);
+ UNION (before, before, tmp);
FREE_WORDS (tmp);
}
return;
--
--Per Bothner
per@bothner.com http://www.bothner.com/~per/