Re: [Nautilus-list] nautilus on freebsd - questions on nautilus_application_startup

[Patrick Hartling <patrick 137 org>]

Darin Adler <darin bentspoon com> wrote:

} On Thursday, July 12, 2001, at 08:41  AM, Klaus Herrmann wrote:
} 
} > i continued my debugging and got a liitle further. nautilus calls
} > Nautilus_Shell_open_default_window(), but never returns from it. as
} > this code is really really ugly autogenerated code, i really don't know
} > what to do next about it. any ideas?
} 
} The ugly autogenerated code is a CORBA stub. This is how CORBA calls work.
}   It should be pretty easy to step through this ugly code (hold your nose 
} while you're doing it) and see where it goes. While the code may be ugly, 
} it's not impossible to decipher. In the end, if it's working properly, the 
} code will end up calling through to a function inside the same process. If 
} you instead find it sending data out a socket or something like that, then 
} you are seeing some kind of ORBit bug, because this should be a call 
} within the process.
} 
} [By the way, the autogenerated code could be made easier to read. But that'
} s a matter for the maintainers of ORBit. I think they are looking for 
} efficiency and are not concerned with how the code looks, but I'd love a 
} debug mode where the code was simpler and easier to read even if a bit 
} slower.]
} 
} > by the way, if i avoid the Nautilus_Shell_open_default_window() call,
} > nautilus complains that it has no viewer capable to display
} > .gnome-desktop - i hope this is a consequence of disabling the default
} > window and not our next problem... :-/
} 
} I expect this indicates a problem with the result of an OAF query, and I 
} presume it will be the next problem. I doubt it has anything to do with 
} the default window issue.
} 
} Have you done any tests of Bonobo and OAF alone? I ask because it's pretty 
} hard to debug ORBit, Bonobo, and OAF using Nautilus as the test bed. If 
} these packages are already known to work, it can save you a lot of time 
} and headaches; it's a lot easier to debug Bonobo and OAF test programs 
} than Nautilus as a whole.

I have some time to try debugging Nautilus 1.0.4 on FreeBSD this weekend,
so I started by trying out the OAF and Bonobo tests.  With OAF 0.6.5, all
the tests passed as long as I started oaf-empty-server before running
'make check' and put broken.oafinfo with the other .oafinfo files.  Is it
normal to need oaf-empty-server for the tests, or should the tests pass
using oafd alone?  I killed oafd every time before running the tests just
to be safe.

For Bonobo 1.0.7, four of the five tests pass without problems.  The
last one, test-xobject, fails when it calls bonobo_object_unref() just
before returning.  Looking through the stack trace, on line 395 of
bonobo-moniker.c, the value of moniker->priv->private is 0xd0d0d0d0 which
is considered a junk pointer.  g_free() fails to release the memory
because of this.

Perhaps this is the wrong place to be reporting these things, but what
I'm interested in is whether or not these two issues could be leading to
problems with Nautilus on FreeBSD?  For the most part, OAF and Bonobo
seem to be working fine.

 -Patrick


Patrick L. Hartling			| Research Assistant, VRAC
patrick 137 org				| 2624 Howe Hall -- (515)294-4916
http://www.137.org/patrick/		| http://www.vrac.iastate.edu/