Re: bonobo activation test segfault on Solaris, forte 6u2

[Michael Meeks <michael ximian com>]

Hi Matthew,

On Tue, 2002-09-17 at 21:52, Matthew Langford wrote:
> Still using forte 6 update 2, moved to bonobo-activation.

	:-)

>  I've set CPP to
> /usr/lib/cpp and recompiled libIDL and ORBit2 2.4.1.  (Could we perhaps
> include a wee note for Solaris Forte users to set $CPP in the INSTALL
> instructions for ORBit2 and libIDL?  In ORBit2, because that's where the
> error manifests itself.  Ideally, I guess, a configure test for this
> would be nice, but that is beyond me.)

	Laca - any chance of a configure check that will test for non gcc
systems, and moan about setting CPP ?

> bonobo-activation 1.0.3 compiled successfully.  "make check" ends with

	Great; note that 'make check' in bonobo-activation is  unlikely to work
for anyone but me :-) it rather requires the stuff to be installed etc.
etc. which is a bit of a pain.

> make[2]: Entering directory
> `/scratch/langfml/bonobo-activation-1.0.3/test'
>  FORKING: '/opt/gnome-2.0.x/libexec/bonobo-activation-server' for
> 'IDL:Bonobo/ActivationContext:1.0Segmentation Fault

	It shouldn't segv though; what trace does the debugger give ?

> Now, trying to do my homework, I found a link from past December about
> changing bonobo-activation/bonobo-activation-init.c to NULL initialize the
> base_service variable.  It looks like that patch is somewhat in place, but
> I fiddle around anyway.  If I change the

	There should be no outstanding fixes for known bugs outside b-a-s CVS.
There is however a worry (of mine) that we're not doing the pre-fork
cloexec thing right for b-a-s at all; which is slightly scary, we need
to fix that really.

	Hmm;

		Michael.

-- 
 mmeeks@gnu.org  <><, Pseudo Engineer, itinerant idiot