🚀 go-pugleaf

>The transport service bridge essentially maps TCP to TP4.  It requires
>that both networks/users/end-points be running the same application
>protocol.  This reduces to requiring that the TCP user also be running
>OSI applications.

True to a point. The topology I was addressing was



   +-------------+  +-----------+ (    ) +-----------+  +-------------+
   |OSI Appl/CLNP|--|trns bridge|-( IP )-|trns bridge|--|CLNP/OSI Appl|
   +-------------+  +-----------+ (    ) +-----------+  +-------------+

    {although RFC 1070 proposes another solution to this problem (CLNP/IP)}
OR
   +-----------+  +-----------+ (      ) +-----------+  +-----------+
   |TCP Appl/IP|--|trns bridge|-( CLNP )-|trns bridge|--|IP/TCP Appl|
   +-----------+  +-----------+ (      ) +-----------+  +-----------+

Transport bridges take care of both of these. Note that with the inclusion
of RFC 1006 (ISODE) a trasnport bridge gives you other connectivity 
possibilities:

   +---------------+  +-----------+ (        ) +-------------------------+
   | OSI Appl/CLNP |--|trns bridge|-(   IP   )-| IP/TCP/RFC1006/OSI Appl |
   +---------------+  +-----------+ (        ) +-------------------------+


The specific point I was responding to was IP<--> CLNP protocol conversion.
I was not responding to any other part of the TCP/IP vs OSI discussion.


>If the two sides must remain "pure" to their base techn>ology -- i.e.,
>there may be no change to the environments on either side -- then you
>are stuck with using application gateways.  These implement all of
>both stacks, all the way up to and including both sets of applications,
>with an "eighth" layer that translates from one application to the
>other.

Yes very true, however IP<-->CLNP won't buy you that, in fact TP<-->TCP
won't buy you that. Let's face it, two machines must talk the same protocol
in order to communicate. If its a layer 7 protocol, then they must talk
the same protocol directly, or somewhere there is an intermediate machine
which acts as an application gateway. I know of no other way.

>Dave

thanks for the input

>bob tausworthe