[omniORB] Object Id and POA reference

Sai-Lai Lo S.Lo@uk.research.att.com
17 Oct 2000 17:49:07 +0100


>>>>> Brenneis, Steve writes:

> Thanks for the quick response, David.
> This leads me to a little more esoteric follow-up. Since the method of
> encapsulating these two components is not specified, would it be allowable
> for an ORB implementation to accomplish this in another way (e.g. private
> tagged profiles)? If so, wouldn't this represent a potential
> interoperability issue since GIOP specifies that private profiles, for
> instance, are allowed, but clients and servers may not rely on their
> presence or absence in any ORB or Request invocation?

To explain this we have to get down to GIOP details.

Up till GIOP version 1.1, the only identification information passed along
in a request is the object key inside the IIOP profile. Therefore an ORB
must be able to decode both the POA id and the object key from this value.

So if an ORB encodes the POA id not in the object key but in some tagged
components for instance, this will not interoperate over GIOP 1.0 and GIOP
1.1. However if the protocol is GIOP 1.2, it is possible for this scheme to
interoperate, although in a very inefficent way. The details is as follows:

In GIOP 1.2, the object identification information can be sent in one of
the 3 forms: object key, a tagged profile or the complete IOR with an index
to specify which tagged profile it is using. One can imagine that an ORB
which encode the POA id in some tagged components inside its tagged
profile(s) would require the clients to contact its servers using either
the 2nd or the 3rd mode of identification. If the client only presents the
object key, the server can response with a reply to ask the client to retry
with the 2nd or the 3rd mode of identification.

On the other hand, it is a spec. requirement that if an ORB supports GIOP
1.2, it should also support GIOP 1.0 and 1.1. In your example, the ORB
surely cannot support GIOP 1.0 and 1.1, therefore I cannot see how an
ORB can be 100% conforming if it does what you suggest.

Regards,

Sai-Lai

-- 
Sai-Lai Lo                                   S.Lo@uk.research.att.com
AT&T Laboratories Cambridge           WWW:   http://www.uk.research.att.com 
24a Trumpington Street                Tel:   +44 1223 343000
Cambridge CB2 1QA                     Fax:   +44 1223 313542
ENGLAND