xml-dev - RE: Services-based automation (WAS RE: Realistic proposals to the W3C?)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

The reliability issue is key.  We did a round on that 
term as a logistics consideration awhile back.  However, 
computing a reliability numbers (eg, Mean Time Between 
Failure, Mean Time To Repairs) and quality assurance 
numbers (Mean Time To Response), we will have multiple 
articles to factor in.  

o  The URI, conflating name 
and location, is problematic.  How can it denote both a 
replicable resource and a bound resource?  That is why 
FPIs and System IDs are separated.

o  The schema.  This has a better chance.  Even if it 
is evolving, when a central authority owns a schema 
and has good change control procedures, the quality of 
it can be controlled.  If it is dynamic in real time, 
all bets are off other than predictive quality.

o  The component.  This has as good a chance as 
testing provides.  The dllHell issue is still with us 
and though the new operating systems are offering 
better options, it comes down to site tuning.  Binding 
it to an authoritative schema improves our ability 
to wire it into the process.  A bad process will still 
be a bad process.

RDF and predicate systems might be one more tool.  
They require yet another skill and toolset 
and with all of the logistics considerations, might 
have a hard time finding a niche among stored 
database procedures, OLAP, and so on.  However, as a 
technology that has not yet emerged with sufficient 
density to characterize as a success, we have to 
take a wait and see stance.   As a basis for a 
vision of the future web, the semantic web seems 
ill-timed and as yet, unrealizable for most 
sites.  A services-based web is a doable, here now, 
must make sure all the pieces work coherently 
kind of vision.  We can explain it, we can implement it, 
and IMO, we can field it reliably.  If we are 
sharp negotiators and professional designers, we 
will get coherence provided the next round of 
standards and specs don't drive us into the weeds.

If we are to offer realistic proposals, we need a 
realistic request for proposals. That's good business. 
If the previous eight years of web emergence can be 
characterized, one might say good splat but bad business. 
That phase is over and we must ensure the standards, 
specifications, and recommendations have a sound 
foundation in the business applications.  

For this, I say services, not semantics.

Len 
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: Gavin Thomas Nicol [mailto:gtn@ebt.com]
Sent: Wednesday, October 18, 2000 9:41 AM
To: xml-dev@lists.xml.org
Subject: RE: Services-based automation (WAS RE: Realistic proposals to
the W3C?)


> Well, DUH! If a semantic web means is a namespace URI points 
> to a schema somewhere, I am completely underwhelmed.

More importantly, this also doesn't work reliably for a
number of reasons... chief amongst them being that URL's
(and URI's in general) are somewhat fragile.