[Css-csts] is text of Framework specification getting enough attention?

[Ray, Timothy J. (GSFC-5830)]

Dear Working Group,

When I was implementing the AMS protocol (for inter-operability testing), a large percentage of my time was spent carefully reading the text in the AMS specification (in order to follow the rules).  While implementing the Proto-Framework specification, only a tiny percentage of my time has been spent carefully reading the text in the Framework specification.  Quite a bit of my time has been spent looking at the ASN.1 source code (trying to figure out how to create the messages).  I'm concerned that the text in the Framework specification isn't getting the attention that it deserves.

There are still some basic protocol rules that don't appear to have been addressed (please correct me if I'm wrong).  For example, there doesn't appear to be any text in version 0.19 of the Framework specification that requires the receiver of a PDU to validate the ProcedureInstanceId.  Suppose an incoming PDU references a ProcedureType that is unrecognized.  That would seem to be a high-level error that would appear in section 3.3.2.7.1 of version 0.19 of the Framework specification - "e) unknown Procedure Type".   Now, suppose the ProcedureType is recognized, but within the ProcedureInstanceId the procedure-number is invalid (e.g. there is one instance of the Association Control procedure running, but an incoming PDU references the 57th instance of the Association Control procedure).  That would also seem to be a high-level error that would appear in section 3.3.2.7.1 of version 0.19 of the Framework specification - "f) invalid procedure instance id-number".

In my opinion, it would be a good idea to specify a sequence of checks that must be performed on every incoming PDU.  It would also be a good idea to specify an additional sequence of checks that are appropriate for each particular type of incoming PDU (Bind-Invocation, Bind-Return, etc) and each particular Procedure.  For example, suppose the Provider receives a Bind-Invocation whose ProcedureInstanceId specifies the BufferedDataDelivery procedure.   I don't think the current text in version 0.19 of the Framework specification addresses this situation.  May I suggest that we add a sub-section wherever these validation checks are required, and that we consistently name the sub-section "Validation of incoming PDUs"?

Best regards,
Tim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ccsds.org/pipermail/css-csts/attachments/20100317/256b28e8/attachment.htm