RE: is that a fork in the road?

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

I am saying you wouldn't have those simple specs if others hadn't 
done the complex ones first.   Inevitable, maybe not but so 
prevalent it must be a pattern.

The first HyTime spec was for a simple presentation 
DTD with a simple goto link in it?  Sound 
familiar.  Some of the first stylesheet driven 
SGML hypertext systems used frames  
similar to DIVs.  Sound familiar?  

As far as I can tell, no one is trying 
to pile 200%.  They are trying to enable 
other stacks to work and haven't quite figured 
out yet where these go into the rock stacks 
because the rocks aren't well color-coded. 
Henry is right about the XML data model.  The 
grove guys were right about SGML.  We don't 
have a firm foundation so every time we 
add a rock, the rest of the rocks start 
to shake (sorry Northwest cats for that 
analogy - I still think Gates should 
have stepped to the mic and asked if the 
Linux contingent was arriving). 

What have we learned from XML's success? 
That the job wasn't done well enough to 
support the follow-on requirements.  

Success?  Everyone can use <...  ...="..." />. 
Whoopee!  Try to connect the dots and see 
what happens.  That minimal victory bites.

Henry's bit is about getting the architecture 
together for the app languages that have to 
operate above the level of bits on the wire. 
I think the complexity we see emerging now 
is because we didn't do that earlier. 

Julius Caesar couldn't take Britain because 
he thought it easy and when it wasn't, it 
scared him back to Rome.  Claudius took Britain because 
he looked at Julius's mistakes and didn't 
make them.  We are on the shoreline.  Are 
we being wise or simply afraid?  One thing 
is certain: we are a long way from riding 
to Rome in triumph.

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

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


-----Original Message-----
From: Simon St.Laurent [mailto:simonstl@simonstl.com]


I hate to point this out, but you seem to have a vision of a world in which 
spec complexity is inescapable, recurring, and inevitable.  I fear you 
don't give communities credit for the potential to learn from the impact of 
past complexities, and expect every new spec to be as hopelessly 
intertwingled as SGML, CALS, and HyTime - _none_ of which qualify as a 
worthy role model for future spec development in my book.

Yes, we need experience to learn which 20% is useful.  We also need to 
foresight to realize that piling 200% on top of the 20% we just slimmed 
down to is probably not going to help much.

>Before we lament the complexity (We are whining!)
>or really sidetrack

I don't understand why you regard efforts to learn from XML's success - 
that doing less is doing more - as whining.

Of course, I tend to regard people who insist that long lists of features 
be piled into what once looked simple and usable as whiners myself, so 
maybe I shouldn't be critical.