Re: Scream and Shout for JacORB



Hi Mark,

On 19-Dec-2001 Mark McLoughlin wrote:
> On Tue, 18 Dec 2001, Roland Juelich wrote:
> 
>>
>> I did an ior-decode, which gives two internet tags with different versions
>> but
>> identical object keys.
> 
>       running this ior through ior-decode-2 doesn't trigger this
> error. Because ior-decode uses CORBA::ORB::string_to_object, the error
> would have been thrown if this was the reference causing the problems.
> 
I just used ior-decode to verify that the object key is equal.

>       The problem reference must come later - are you returning an
> object from one of your calls? The best way to find what's going on
> might be to set a breakpoint in g_log and get a backtrace from there.
> 
I have an interface as parameter in the call that produces this error. BTW, the
warning occures twice, and I have now found, that a method call to a function
of the interface fails. The details:

IDL contains:

        ...
        interface ServerCB
        {
            oneway void handleEvent(in Events events );
        };
        ...
        interface Device
        {
            void register_event_callback ( in ServerCB eventCB )
                raises ( thema::TheMaException );
        }

For interface ServerCB, ORBit is client, for interface Device, ORBit is server.
The error occures, if register_event_callback() is invoked by the JacORB client.

Then a following call to handleEvent produces a JacORB exception:

[ POA TheMaRegistration - rid: 2147481680 oid: ServerCB-1008768501906 opname:
handleEvent - invocation: system exception was thrown
(org.omg.CORBA.BAD_OPERATION: handleEvent not found  vmcid: 0x0  minor code: 0 
completed: No) ]

I'm not sure wether this JacORB exception is caused by the preceding ORBit
warning (error). What do you think? I do not know, what consequences the error
ORBit internally has.

>> > Apart from this warning did you have any other problems?
>> I had to eliminate the multiple component tag with the TAG_CODE_SETS, which
>> caused exception in JacORB.
>>
>> > Does the ORB function even *with* this warning?
>> >
>> I reduced the error to a warning and than this part is working fine.
>> Simple functions are also working, but I still have a core dump on ORBit
>> side. I
>> assume there is still an alignment problem. I will have a look tomorrow.
> 
> Could you send on more details of the other problem when
> you have a chance. I fixed some issues when running between big/little
> endian machines, but there may be more issues lurking.
> 
It is fixed. Cause I crosscompile to powerpc, I have to set the alignments by
hand (configure does not work on that topic if crosscompiling) and handwork is
faulty ;-)


Roland

--
ATecoM GmbH                              Date: 19-Dec-2001
Kaiserstrasse 100                        Time: 14:23:37
D-52134 Herzogenrath

E-Mail: Juelich@ATecoM.com
Tel.:   +49-2407-9586-0
Fax.:   +49-2407-9586-901
--------------------------------------------



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]