RE: The relentless march of abstraction (fwd)

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

Probably because we still have some practicing 
to do designing coarse transactions.  It isn't 
easy or straightforward to take previously short 
fast transactions and glomming them to pass the 
ACID test.  We should discuss that more.  Are 
any best practices emerging yet?  Do we need 
conceptual design tools for that?   Also, 
there aren't enough clients capable of in a 
shared and reliable way doing the right thing with 
aggregate docs and federated pages.   A lot 
of folks are still designing to BasicHTML 
paradigms for statelessness.   

Really, I don't think the SunGURU position 
of fat server/thin client or the MicrosoftGURU 
position of fat client/fat server works for 
all cases.   A message switch is the crux.

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

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


-----Original Message-----
From: Tim Bray [mailto:tbray@textuality.com]

At 11:41 AM 27/02/01 -0500, David Megginson wrote:

>I think that client-side XML failed simply because it didn't fill a
>big enough real need (HTML 4 is close enough)

I have a problem with your verb tense.  The web is still too slow.
Fatter pipes aren't going to help.  The only way to make it fast
is to do some of the work on the (severely underemployed, these
days) client, and the only way to do that is to send some useful
data there to get chewed on.  So I think client-side XML just
hasn't got going yet.  To say it had failed, it would be 
necessary for it to have been tried. -Tim