[BYTEIO-WG] ByteIO Interop Document

[Stephen M Pickles]

Mark,

I agree with you.

As I see it, the objective of your interoperability work is
in order to validate your own specification, not to test the
interoperability of specifications (and their implementations)
on which you depend, such as WSRF. Thus, I would say that the
following questions are in scope:
1) is the specification sufficiently clear and free of ambiguities
   to allow the independent development of interoperable
   implementations?
2) do the interoperability testing experiences highlight a need for
   clarification, such as a profile of how dependent specifications
   should or should not be used to achieve interoperability?
3) do the interoperability testing experiences expose a deficiency
   in the spec itself which should be addressed by errata or
   another revision?
And, IMHO, the following can be ruled out of scope
(bring them into scope at your peril):
1) proving that a given set of implementations do indeed interoperate
   (the main business of OGF is standards, not implementations.)
2) testing the interoperability of dependent specifications.
3) testing the compliance of an implementation with your specification
4) testing the compliance of an implementation of a dependent
   specification.
(This is not to say that all these aren't important - one hopes that
sooner or later all of these do get addressed. But the working
group shouldn't have to take them all on.)

This is easy to say, but can be much harder to do. My advice,
FWIW, is to be pragmatic - do what you reasonably can to achieve
your objectives without opening too many cans of worms, and
document the rationale behind key decisions along the way.

I'm cross-posting this to the interop-bof list, in the hope that
others will have constructive comments or relevant experience to
share.

Best regards,

Stephen

> -----Original Message-----
> From: byteio-wg-bounces at ogf.org 
> [mailto:byteio-wg-bounces at ogf.org] On Behalf Of Mark Morgan
> Sent: 03 October 2006 16:32
> To: 'Mailing List for BYTEIO-WG'
> Subject: [BYTEIO-WG] ByteIO Interop Document
> 
> Ladies and Gentlemen,
> 
> On the telecon today, an issue was raised about the interoperability
> document that needs to be addressed this week if possible 
> (via email if
> possible).  Basically, everyone seems to agree that we need 
> to get WSRF
> ResourceProperties to test the properties that part part of the ByteIO
> specification (getting the size of a ByteIO resource, etc.).  
> However, I
> personally am of the opinion that we should minimize the amount of
> non-ByteIO specific mechanism that we test as our purpose is 
> only to test
> ByteIO and not WSRF.  That said, I think we should utilize 
> the simplist
> Resource Property operation, namely GetResourceProperty, but 
> not use other
> WSRF-RP operations such as QueryResourceProperties and
> GetMultipleResourceProperties.  I just don't feel like their 
> inclusion in
> our test serves any purpose beyond interoperability testing 
> of WSRF which
> isn't our goal.  Thoughts?
> 
> -Mark
> 
> --
> Mark Morgan
> Research Scientist
> Department of Computer Science
> University of Virginia
> http://www.cs.virginia.edu
> mmm2a at virginia.edu
> (434) 982-2047
> 
> --
>   byteio-wg mailing list
>   byteio-wg at ogf.org
>   http://www.ogf.org/mailman/listinfo/byteio-wg
> 
>