Re: Web Object Services

At 11:21 PM 2001-09-10 , Brendan Macmillan wrote:
>You might also contact some of the existing standards bodies - they certainly
>will be motivated by many of the same issues as your clients!

Please add 'emergent' consensus-seeking dialogs.  In other words, add the
Global Grid Forum.



This is in some sense a niche community, heavily committed to some lagging

On the other hand a) they have problems that genuinely require a high quality
of coordination amongst computations executing on largely autonomous platforms
and b) they do a lot of this.  These computations are expensive so there is
valid economic energy behind the proposition to make the whole thing work
better.  And c) being heavily government and academia based, they will talk
more openly about what they are doing than some others also investing in
mounting web services.

From my role as advocate for flexi-mode service delivery with regard o user
interface particulars, 

c.f. <>

I see the Web today as more affordable but less functional that an full-up
distributed object world as I think is envisioned by Jim.  And some of that
functional disadvantage hurts my clients.

Through the WAI in the W3C arena, and through the PACI UD/DA program in the
Global Grid Forum, some of us are seeking the right small set of demands on
way Web services are implemented in general so that a reasonably equal
level of
service is achievable by readily available user-side-led-and-driven
of the service delivery channels.  The Computational Science community that
GGF serves as customers appears to need a lot of the same consistent quality
delivered over diversity of delivery channels as the people with disabilities
using or not using assistive technology as they may be.  That is why I regard
GGF and their clients in Computational Science as allies in the pursuit of
consistenly high-quality results from flexibly web-mediated service delivery

For a soundbite, one could say that "what people with disabilities need from
the W3C is just that they get serious about Device Independence."  And the
people who think they need Grids need a lot of the same things (from the Grid
Services consensus core of practices).

In the process, we will be stupid not to carfully mine the lessons learned
work on distributed-object worlds.  But it is unlikely that the solutions will
be taken over wholesale,  The only problem with object-oriented approaches is
that they wind up _too_ narrowly fixated on the individual objects.  Viewing a
method access on a remote object as a service rendering transaction, and
including services accessed verb-object as well as these services addressed
object-verb is a healthier omnibus view.  See the <> archives for my
many rants about how search URLs are verb-only service requests, and the
service rendered is to identify the objects.  Just given the passage of time
and the gradual consolidation of lessons learned, there will probably be a
better global price/performance balance deal struck in the end for all.


for more on this.


Received on Tuesday, 11 September 2001 09:36:03 UTC