OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] Competing Specifications - A Good or Bad Thing?

[ Lists Home | Date Index | Thread Index ]

From: Rick Marshall [mailto:rjm@zenucom.com

>my beef with all of this is simple:

>1. a belief in the necessity of anarchy. and most standards efforts are
>directed against the intrusion of anarchy, rather than the utilisation
>of it.

To me, that's serendipity.  Anarchy is just lawlessness or lack of a 
working address.   Chaos is the engine of evolution.   We don't have 
to create chaos; it is perceived.  (all politics are local).

>2. objection to commercial interests and marketing directing technical
>direction. doesn't it worry anyone else that these standards are driven
>by ibm, bea, microsoft, sun etc not prof this and dr that? that as a
>result they can only be implemented at great cost by large teams in big
>products, and everyday software developers have no hope?

That worries me least of all.  IBM, BEA, Microsoft and Sun have to respond 
in near real time to real requirements from real customers.   We may not 
like that, but that is where the chaos comes from, so if you are a lover 
of anarchy, the multiple inconsistencies from real time customers feeding 
back through multiple vendors independently trying to smooth them out 
guarantees it.  On the other hand, I sit sometimes and read these articles 
from Profs and Docs that regurgitate what the commercial market already 
knows and laugh out loud.  I think commercial apps and university research 
can make handsome couples if they don't make out in public too much.

>3. not particularly committed to open source, but on average it has done
>more good than bad for me.

Open source is a good thing, IMO.  People who believe it to be a weapon 
against enemies of their own choosing don't do it much good.  People 
who believe it to be an enemy don't do themselves much good either.
Spy Vs Spy.

>4. my problem with microsoft is that it is as much a conservative force
>slowing computing to a speed that it can exploit for maximum profit.
>bill gates claims notwithstanding, i don't feel that computing is very
>much advanced on where it was in the 80's. it's just more ubiquitous (is
>that a tautology?)

There are POV issues there.  There are those who say the open source 
products that make installation for the average buyer impossible or 
very difficult are holding back progress.  And, yes, slowing down 
the integration of innovation is pretty common in every industry. 
Mercedes Benz, for example, refuses to add any feature to a car 
that isn't completely and testably reliable.  So they build great 
and expensive cars but are never at the forefront of innovations 
even if they invented them.   There are those who claim rightly 
in my opinion that without the ubiquity of the MS desktop, Mosaic 
and the WWW would still be used mainly by Docs and Profs and students. 

In other words, computing advances at an overall slow rate.  The 
picture to keep in mind is lots of little systems all evolving at 
different life cycles.  MS happens to be a big one, but if you 
isolate them, you again see, lots of little systems evolving 
at different life cycles.

>5. i still develop lots of small interacting processes, even on the one
>machine. distributed processing should just be an extra communication
>layer, not a whole "new" way of doing things.

Sure.  The different approach is to rely on a common syntax and shared 
document definitions instead of types and operators.  RPC didn't work 
at scale.  Scale is the issue, I think.  If you are local, why 
bother with the cumbersomeness of what works out to be the largest 
document model ever attempted.

>6 having said that the only bits of clear interest to me in the whole
>ws-* group are 1. reliable messaging (in this case a rework of things
>we've done a dozen times before), 2. reliable identification (also a
>rework), 3. encryption for secure communications (also a rework).

>and if you look at the bits that work and are starting to be used widely
>it's just those things. 

So far so good.  Above that, as I read so far in the MS document, you 
only need the rest when you have to create webs of federated trust and 
reliabilty and the illusion of seamlessness for an end user when 
incorporating services made by people you don't know and don't trust.

It's like rental cars: you want them to work until you drive them back 
even if you abuse them every mile of the trip.



News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS