[omniORB] stuck with python string size

[Joost van Lawick]

On Thu, Jan 30, 2003 at 11:07:19AM +0000, Duncan Grisby wrote:
> On Tuesday 28 January, Joost van Lawick wrote:
> 
> > I use omniOrb and Python to connect to a Visibroker orb. I have
> > compiled the latest omniOrb and omniOrbpy from CVS on a (suse) linux
> > box. I have added --enable-threads and --disable-static to both
> > configure scripts. I can connect send and receive python strings but
> > only of size < 1024 chars.
> > 
> > When I send a message (size >= 1024 chars) I get the following message:
> > 
> > -----------------
> > omniORB: Client opened connection to giop:tcp:192.100.100.243:51566
> > omniORB: sendChunk: to giop:tcp:192.100.100.243:51566 112 bytes
> > omniORB: sendCopyChunk: to giop:tcp:192.100.100.243:51566 10288 bytes
> > omniORB: sendChunk: to giop:tcp:192.100.100.243:51566 43 bytes
> > omniORB: throw giopStream::CommFailure from giopStream.cc:816(0,MAYBE,COMM_FAILURE_WaitingForReply)
> 
> It looks like perhaps Visibroker doesn't support the fragmented
> messages that omniORB is sending. If that is the case, it's broken.
> To find out, you could try setting omniORB's maximum GIOP version to
> 1.0 with -ORBmaxGIOPVersion 1.0 .
> 
> If that doesn't help, is there any tracing you can turn on in
> Visibroker?
> 
> Cheers,
> 
> Duncan.
> 
Hello Duncan,

Thanks for your reply. Setting maxgiop to 1.0 does indeed solve the problem. I
also found other omniOrb - VisiBroker related problems in the mailing archives
and one in particular mentioned the default backward compatibility of
Visibroker 4.5 which causes problems with the length of null strings. I was not
able to test this so far but maybe my original problem disappears if I can turn
off Visibrokers backward compatibilty. I will keep you posted. For the moment,
GIOP 1.0 will do.

Cheers,

Joost

-- 
Once, adv.:
	Enough.
		-- Ambrose Bierce, "The Devil's Dictionary"

--
Joost van Lawick van Pabst
E: joost@lawick.nl
W: http://www.lawick.nl/