[ogsa-d-wg] Format for interface descriptions in OGSA-D

[Stephen Davey]

Hi Michael,
I prefer your version for the format for interface descriptions. Though
it is perhaps not ideal to include the actual operation and property
names, it does make the interfaces much more readable. Adding the table
lines will also make them stand out more, which will help me when I try
to match up the scenario steps with appropriate interfaces - whether we
want to emphasize the interfaces in the data architecture document is
another matter I guess.

Cheers, Stephen.


-----Original Message-----
From: owner-ogsa-d-wg at ggf.org [mailto:owner-ogsa-d-wg at ggf.org] On Behalf
Of Michael Behrens
Sent: 23 December 2005 02:08
To: Dave Berry
Cc: ogsa-d-wg at ggf.org
Subject: Re: [ogsa-d-wg] Format for interface descriptions in OGSA-D

I attempted to rearange the text a bit and added lines, thinking that 
might be easier for the reader when confronting that pattern?  It's 
attached - not sure if you'll like it.  Its a little more UML-ish.

BTW:  The strawman architecture (v601) is looking good - covers a lot of

ground!   Good concepts for global access to information via caching, 
naming, etc. 

Dave Berry wrote:

>Folks, 
>
>On yesterday's telcon we discussed how to describe interfaces in the
>OGSA Data Architecture document.  I undertook to draft a straw man
>description of the ByteIO interface as an example.  I attach this draft
>for comment.  (Note: it is based on the ByteIO interface described in
>our document, which is significantly out of date.  What we're
interested
>in here is the format, not the content).
>
>On the telcon we agreed that we wanted to abstract away the details of
>the operations such as their names and the names and types of their
>parameters.  This is so that (a) we don't swamp our readers with
detail,
>and (b) we don't have to keep changing our document to track changes in
>the detailed specification.  On the other hand, we do want to describe
>the key operations, so that (a) readers can better understand what the
>interface does, and (b) we show how the interface will interact with
>other parts of the system.
>
>I'm not wedded to the format in the attachment.  It is purely intended
>as a starting point for discussion.
>
>Best wishes,
>
>Dave.
>  
>

-- 
Michael Behrens
R2AD, LLC
(571) 594-3008 (cell) *new*
(703) 714-0442 (land)