Ben Greear <[EMAIL PROTECTED]> wrote:

> Pavlin Radoslavov wrote:
> 
> > Please apply the patch below and see whether it solves the stale
> > memory problem and/or some of the earlier problems.
> 
> I took a different path but my fix is somewhat similar.
> 
> I'm tracking down some more memory leaks found by valgrind
> and hope to have a patch for review within the hour.  I'll
> post my patch, but if you prefer your method (and it works),
> then that is fine too.

OK, I will wait for your patch and compare it with what I have
before committing anything.

> Did you get a chance to look at the bug_catcher logic?  Any
> interest in adding that?  I think it will be useful for
> catching future bugs & debugging core files.  It's especially
> nice for core files because you can print out the 'magic' member
> and see if an item has been deleted or not....

I looked into it, but the downside of a solution like this is
you have to modify the source code in the exact location you think
there is a problem. Also, the only problem it seems to solve is
deleting an object that has been deleted previously.
This obviously makes its applicability very limited and I am not
sure how much effort will be saved by using it.
In fact, given that it typically requires modification of the *.hh
header file with the class declaration, it might actually slow you
down waiting for everything that depends on that header file to
compile.

Myself I am using similar hacks from time to time, but typically
few simple "XLOG_INFO()" print messages are sufficient to tell me
whether I am looking into the right direction.
If it seems the problem is memory-related, then tools like valgrind
should be used instead.

Said that, I don't think we should push it into XORP because it
might be of little value to other developers.

Thanks,
Pavlin

_______________________________________________
Xorp-hackers mailing list
[email protected]
http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-hackers

Reply via email to