Clarifications on ESRO SAP Addressing

[Mohsen Banan <mohsen@neda.com>]

>>>>> On Mon, 18 Aug 1997 08:17:17 -0400 (EDT), Scott Bradner <sob@harvard.edu> said:

  Scott> a comment from one of the IESG members about your draft

  Scott> Scott

Harald's observations are correct.

That part of the spec is not clear.

I will fix this problem in the next rev of the spec. 

My responses to  Harald's email follow:

  >> From hta@dale.uninett.no Mon Aug 18 08:05:50 1997
  Harald> X-Mailer: exmh version 1.6.9 8/22/96
  Harald> From: Harald.T.Alvestrand@uninett.no
  Harald> To: Scott Bradner <sob@harvard.edu>
  Harald> cc: iesg@ietf.org
  Harald> Subject: Re: draft-rfced-info-banan-esro-00.txt
  Harald> In-reply-to: Your message of "Sun, 17 Aug 1997 20:48:03 EDT." <199708180048.UAA08679@newdev.harvard.edu>
  Harald> Mime-Version: 1.0
  Harald> Content-Type: text/plain; charset=us-ascii
  Harald> Date: Mon, 18 Aug 1997 12:47:04 +0200
  Harald> Sender: hta@dale.uninett.no

  Harald> The broken part of the protocol is this line:

  Harald> 4.6.3  Use of lower layers
  Harald> 
  Harald> ESRO protocol uses UDP port number 259.

Agreed.

Section 4.6.3 needs to be expanded to explain that the ESRO SAP
Selector's range is only 16 (4 bits). That the assignment of a UDP
port is part of the application layer. That port 259 (esro-gen) has
been assigned for general usage of esro.

  Harald> The protocol is an obvious alternate to "heavier" RPC protocols,

Glad you feel that way.

  Harald> but has only 8 bits of SAP to differentiate between different users.

It is not 8 bits for the ESRO SAP Selector.
It is only 4 bits (a total of 16).

The design decision to accept a limited ESRO SAP Selector space for
efficiency and to rely on UDP ports for wider use was deliberate.

  Harald> The "obviously right" thing to my mind is to say that the assignment
  Harald> of an UDP port is part of the application protocol.

Agreed.

  Harald> The document is also self-conflicting; see this paragraph:

The conflict is because section 4.6.3 needs to be expanded.

  Harald> 2.4.2  Performer-address

  Harald> This parameter is the address of the ESROS Performer User which 
  Harald> consists of ESRO Service Access Point (SAP) Selector, Transport 
  Harald> Service Access Point (TSAP) Selector (e.g., port number), and Network 
  Harald> Service Access Point (NSAP) address (e.g., IP address).  This 
  Harald> parameter has to be supplied by the invoker of the service.

  Harald> ESROS Invoker User provides the Performer-address parameter for the 
  Harald> ESROS-INVOKE.request primitive. 

  Harald> which seems to call the port part of the address.

Section 2.4.2 is correct. Section 4.6.3 needs clarification.

  Harald> I have no objection to publication, but if you talk to them again,
  Harald> it might not hurt to clarify this part.

Thanks for your comments.

...Mohsen