This is the mail archive of the java@gcc.gnu.org mailing list for the Java project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Tree check failures when compiling Classpath


Hi,

On Wed, 2002-03-13 at 18:00, Alexandre Petit-Bianco wrote:
> Interesting. I'll be trying to look into that soon. Is there a PR
> already filed for this?

Just filed it and made sure that it can easily be reproduced with the
latest stable GNU Classpath release (0.03).
<http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&database=gcc&pr=5942>

I found this by doing the following. I assumed that current_class should
always contain a real class and never a expr_with_file_location. So I
made the following define

/* Debug sanity check */
#define TYPE_CHECK_CURRENT_CLASS \
  if (current_class) {TYPE_CHECK(current_class);}

And added a TYPE_CHECK_CURRENT_CLASS after every assignment in parse.y.
Then you can run gj1 under gdb, break on tree_class_check_failed and
check with print debug_tree what the actual tree node contains.

I still don't really understand the code, but at least I found this.
Is my assumption about current_class correct?

Cheers,

Mark


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]