Re: "Conditions" in the Conformance White Paper

[Lynne Rosenthal <lynne.rosenthal@nist.gov>]

Todd - 

Thank you for your comments and suggestions.  
>  
> Re: Appendix C, where you offer an
> example, section C.4.1 below item (2).  You list the conditions that
> a Registry and Registry Client must satisfy for conformance.  Since
> this is an example test plan, I would like to see a mapping of those
> conditions back to the specification.  For each condition, I would
> find it useful to identify the source statement in the spec that caused
> the condition to exist.  

It is always good to have some traceability back to the
specification.  Since these clauses reference the registry
specification, it probably makes sense to include the clause number from
the specification.  
>  
> In general, is there a formal (or
> defined or recommended) methodology for reading a specification and
> identifying conformance conditions? 
Good question.  I don't know if
there is something. I think usually someone reads the specification and
looks for words like SHALL or MUST.   However, we (the
Conformance TC) is in the process of creating a Conformance Requirements
document which may help.  Within this document is a section that
identifies the conformance keywords that are used by ISO and those use by
the IETF.   We plan to circulate a draft of this Conformance
Requirements document prior to the telconf on Aug 16. 

>  
> Related to this, your paper talks
> about disputes in the testing process.  What about disputes that
> might arise in deriving (agreeing on) the conditions for
> conformance?  Is this even an
> issue?
Yes, this is sometimes an
issue.  It often occurs and is a discussion when developing the
conformance clause for the specification.  That is one of the
reasons we think having a conformance clause in the specification should
be mandatory.  By having this clause, it minimizes the disputes over
when is something conformant, later on.  
>  
> Finally,  I am thinking it would
> be good to have a formal definition of "condition" and then
> adding it to you glossary. I'm not sure of the best phrasing for this,
> but the context for the definition would be: "an implementation is
> conforming if it meets the following conditions:"

Good idea. We will have to work on a good definition

regards.
Lynne

Lynne S. Rosenthal
NIST, 100 Bureau Drive, Stop 8970
Gaithersburg, MD 20899-8970
(301)975-3353, fax (301)948-6213
lynne.rosenthal@nist.gov