[omniORB] Visibroker with [J]C++Builder

[mik@geosys.ru]

Hello, Duncan!

27.04.01 15:32:53, Duncan Grisby <dgrisby@uk.research.att.com> wrote:
>> First, the interopearble speed was about two times slower than in
>> pure VisiBroker/VisiBroker or omniORB/omniORB configurartions
>
>Are the client and server running on the same machine? 

Yes, really.

>VisiBroker uses a shared memory transport in that case. When talking
>to omniORB, it would have to resort to TCP loopback, which would
>account for the slow-down. According to some performance measurements,
>omniORB talking to itself through the TCP loopback is faster than
>VisiBroker talking to itself through shared memory.

Will a shared-memory version of omniORB (4?) be even faster?

>> Second, while omniORB was able to talk to VisiBroker's Naming
>> Service, vice versa access failed - specifically registering new
>> names from VisiBroker client with omniNames.
>
>In what way did it fail?

I tried to reproduce the situation and found a working combination which something puzzles me. Everything - omniNames, VisiBroker-based server and omniORB-based client - 
run on the same machine. Both server and client look for [corbaname:iiop: | iioploc://]127.0.0.1:3001/NameService (that is configured with proper command-line and 
configuration-file options, tested). And here goes the difference:

With VisiBroker, I start "nameserv NameService -J-Dvbroker.se.iiop_tp.scm.iiop_tp.listener.port=3001" (no host specified) and everything works fine for both local and remote 
access to this Naming Service.

With omniNames, I have to explicitly run "omniNames -start 3001 -ORBInitialHost=127.0.0.1" (option -start for the first run only) to make everything work locally.

Is this all right?

Best regards,
		Mikhail

Dr. Mikhail Soukhanov <mailto:mik@geosys.ru.>

Laboratory of Geoinformatics, VNIIgeosistem
Warszawskoje chaussee 8, Moscow M-105, 113105 Russia
Tel.: +7(095) 954-21-50 (x101), fax.: +7(095) 958-35-22
W.W.W.: <http://www.geosys.ru./>