Re: [Fwd: Success with Nutch & GCJ]

[Hans Boehm <Hans dot Boehm at hp dot com>]

I presume you found GC_set_warn_proc()?  It's possible to do all of
this outside the collector.

I'm not completely sure what's best here.  This might be a useful option for
the collector in general.  But it's ability to generate stack traces is
probably considerably more limited than gcj's.

Is the Google code limited to X86?  I probably wouldn't want the collector
itself to depend on it anyway, though it does sound very useful.

Hans

On Tue, 14 Feb 2006, Anthony Green wrote:

> On Tue, 2006-02-14 at 13:24 -0700, Tom Tromey wrote:
> > >>>>> "Hans" == Boehm, Hans <hans.boehm@hp.com> writes:
> >
> > Hans> You might be able to determine who is doing the allocation by setting a
> > Hans> break point in GC_default_warn_proc, and looking at the call
> > Hans> stack.
> >
> > Maybe we could have that (optionally) print a stack trace for us.
>
> I get this message from azureus as well and was planning on having it
> dump core instead:  http://goog-coredumper.sourceforge.net/
>
> I've tried the google coredumper, and it really works.
>
> AG
>
>
>