By the way, one thing I noticed from your previous mail, you start the
ab with the command line
ab -ORBInitRef NameService=corbaloc::192.168.0.102:12019/NameService \
-WATMGIOPAddr inet:192.168.0.102:12010 \
-WATMGTPAddr inet:192.168.0.102:12020 \
-WATMBridgeName CarlosBridge --POAImplName MobilitySupport -ORBNoIIOPServer
> ab.ref
There should be only one dash in '-POAImplName'. This shouldn't be
the cause of your problems, but I'm not sure whether it can cause
other problems.
"Carlos Eduardo" <[EMAIL PROTECTED]> writes:
> I've attached a file that configures everything that I need, don't be
> scared with the code, it works, but in this "beta" version do some
> errors, like this one. The messages are in Brazilian Portuguese
> language, but the variables are in english, is a Shell Script file.
There was no attachment with your message. Did you forget to include
it? Anyway, if my suggestion below works, it's not needed.
> Here is the function called, in the "bridge.cc".
> The error is pointed:
> ...
> void WATMBridge::GIOPBridge::register_local_object (CORBA::Object_ptr o)
> {
> CORBA::Long len;
>> const CORBA::Octet *pkey = o->_ior()->active_profile()->objectkey(len);
> CORBA::OctetSeq key(len);
> key.length(len);
> memcpy(key.get_buffer(), pkey, len);
> _local_keys.insert(key);
> }
Okay, I'm pretty sure I know where the problem is. In bridge.cc
there's the following code starting from line 79 in the GIOPBridge
constructor:
_orb->ior_template()->add_profile(prof);
_tservers.push_back(tserv);
Between these two lines, add the code
if (i == 0)
_orb->ior_template()->active_profile(prof);
It seems like this should work. However, I'm a bit puzzled as to why
I never saw this problem because it should manifest itself on every
run, and I did test the wireless parts with 2.3.12 at some point.
Karel, if you're following, did something change recently in how the
IOR class handles its active_profile member? It seems like
del_profile will set it to NULL when it's deleted, and add_profile
does not set it even when it's adding the first profile. This seems
to me to be the cause of this problem, since a MIWCO bridge starts by
deleting all profiles from the IOR template.
> There is a "Try catch" in the first code, and the error is inside the
> "try", so if it got an error, the logically thing to do is "catch" the
> error, and go ahead, and do not stop the program..
> So the error is still there.
The try-catch doesn't matter. A segmentation fault is a signal, not
an exception.
--
Jaakko Kangasharju
Operating systems as geopolitical areas:
Debian is China
_______________________________________________
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel
