[omniORB] Using omniNames with Visibroker 4.0

[Tamas Morvai]

Hi,

I get a segmentation fault when my server application (Visi 4.0)
wants to bind an object to the omniORB 3.0.2 Naming Service.

I started omniNames with the following parameters

  -start -logdir /home/omniorb/ -ORBInitRef
NameService=corbaname::localhost

and I started the server application with the following
parameters:

  -ORBInitRef NameService=iioploc://localhost:2809/NameService

and I get the following core dump:

[1] VISGIOPProtocolConnector::_reset_object_key(0xef5bbdc4, 0x1,
0x131234, 0xef5dcdfc, 0x130b14, 0x1311e8), at 0xef352990
[2] VISGIOPProtocolConnector::VISGIOPProtocolConnector(0xef5bbdc4,
0x1311ec, 0x1311f4, 0x1311fc, 0x131204, 0x13120c), at 0xef351ab0
[3] VISIIOPBidder::Bid::getConnector(0x0, 0x0, 0xefffeac0, 0x1311e8,
0x1311e8, 0x1311a8), at 0xef345c5c
[4] VISPortfolio::getConnector(0x0, 0x0, 0x0, 0xefffeb2c, 0xef574bf0,
0x0), at 0xef346d04
[5] VISProtocolManager::getConnector(0x0, 0x0, 0x10a380, 0xefffeb9c,
0xefffebb4, 0x12e040), at 0xef34a62c
[6] VISDelegate::_bind(0x130b28, 0xefffec60, 0xefffec78, 0xef5bbdc4,
0x130b58, 0x130b40), at 0xef34cdec
[7] VISDelegate::_verify_connection(0x130b28, 0x12f7fc, 0x1, 0xef5bbdc4,
0x1, 0xfffffb2d), at 0xef34ee18
[8] VISDelegate::request(0x1, 0x12f7fc, 0xef7946c4, 0x1, 0x130b28,
0x12f7fc), at 0xef34d544
[9] CosNaming::NamingContext::rebind(0x12f808, 0xefffedec, 0x12df54,
0xef7946c4, 0xefffede8, 0xef56e9f8), at 0xef784e88


The same server application works with Mico's Naming Service.
I started Mico's Naming Service with the following parameters:

  -ORBIIOPAddr inet:localhost:2809

When the server application is up then I can send messages to it
with a client application which uses omniORB.

Of course, if I compile the server app with omniORB then
it works with omniNames.

I'm using Solaris 2.6 and I compiled omniORB 3.0.2 with gcc 2.95.2

Can anybody help me with it? It is a bug in omniORB or in Visibroker
or am I doing something wrong?

/Tom