Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!CS.WISC.EDU!hagens
From: hagens@CS.WISC.EDU
Newsgroups: comp.protocols.iso
Subject: Re: ISO Session Help!
Message-ID: <8908181607.AA05269@janeb.cs.wisc.edu>
Date: 18 Aug 89 16:07:35 GMT
References: <4756@ubc-cs.UUCP>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 38


> >
> >   Whenever a category 2 SPDU is sent will a category 0 SPDU
> >   ALWAYS be concatenated onto the front?

>     I read the CCITT 1984 X.225 Recommendation. For concatenation, a
> category 0 SPDU is always in front of category 2 SPDU(s). As for the
> valid basic concatenation of DT SPDU, the first pdu is GT SPDU ONLY.
> PT is not allowed with DT SPDU. There is some rule about this GT. Token
> item parameter is only present in the GT SPDU if this DT SPDU contains
> a complete SSDU or the last the segment of a segmented SSDU.  (Pages
> 315-318 of the X.225 document give a good description about valid
> concatenation)
No!
The GT SPDU is always prepended to a DT SPDU. The token item parameter
is present in the GT SPDU iff the ss-user wishes to transfer tokens.
If the token item parameter is not present, then the GT has no effect (although
it still takes up 2 bytes.

> >Can you ever have a TSDU with JUST a data SPDU contained in it?
If you mean, just a DT SPDU, the answer is no. You always concatenate.

>     A DT SPDU can be mapped to a TSDU, but it would be considered to
> be an invalid PDU by the receiver.
Correct. Don't do this.

> QUESTION for the ISO Session Layer gurus:

> Why are some SI values of several SPDUs the same (e.g. MAA and AEA)?
> If the purpose of SI values is to identify the SPDU type, shouldn't 
> they be all unique? It would be a lot easier to decode a SPDU had the 
> SI values been all unqiue. 
The session protocol was designed to be bit compatable with a pre-exsistent
protocol (T.61). This ambiguity of SPDU SI values is a result of this. 
It is possible to distiniguish the PDUs by looking at the parameters that
are present.

Rob Hagens