[
Lists Home |
Date Index |
Thread Index
]
The problem, Rick, is they sometimes do use those features in public
specifications with a lot of reach. See GJXDM.
Otherwise, yes: let the local requirements rule the choice and over
time, some will converge as long as the environment remains stable.
However, it might not. That is the challenge of the low energy transfer
means as described: the superhighway keeps moving because the masses
that create gravity wells do.
(hmmm, that is a lot like the dormitory stairwells at Hogwarts.)
The reason many complex standards didn't get traction is not because of
complexity only but because of the mapping of complex feature sets to
dynamic environments. If you look at the famous examples
(HyTime/DSSSL, OSI, CORBA, etc.) in the context of the time in which
they were attempted, you see not only are they hard to implement and
understand, they are promoted at the time that the reach they have to
make grows dramatically. Given the Long Tail effect in which only a few
have the necessary background to choose wisely, the mean falls into a
subtoptimum minima: mediocre. It isn't that worse is better or
Internet Time means results must be fast, it means one has to sell to
and implement for a thin but wide locale. Intelligence doesn't scale.
The web is mediocre. As long as one thinks 'standard' equals
'universal', these spat continue just like a Highlander movie. When
one begins to grasp the unavoidable nature of situation semantics and
costs, one begins to take a navigational view over time and a changing
landscape.
It isn't a web of objects; it is a web of force vectors over object
evolution. It isn't unpredictable; it is navigable where cost equals
speed and the parametric commodity is time.
len
From: Rick Jelliffe [mailto:rjelliffe@allette.com.au]
The RELAX NG versus XSD competition is bogus, consequently treating one
as the winner and one as the loser at this stage is bogus.
They are both grammar languages, and in many cases you can substitute
one for the other. In fact, it seems that people rarely use the features
in one that are not also not in the other, at least for public schemas.
See
http://www.oreillynet.com/xml/blog/2006/04/reconciling_xsd_and_relax_ng.
html
for a reference to Expressiveness and Complexity of XML Schemas
<http://alpha.uhasselt.be/%7Elucg5503/tods2006.pdf> by Martens, Neven,
Schwentick and Jan Bex which has a lot of material of interest. (You
might even interpret that survey as saying that people are *not* buying
into the exotic parts of XSD. I'd expect the same of RELAX NG.)
People using DTDs, XSD or RELAX NG are all buying into grammars, with
all that entails. I know I have a much more XPath versus grammar
viewpoint than many people, but I think grammars, whether DTD, XSD or
RELAX NG, will play a decreasingly important part in document processing
(perhaps XQuery/XLinq/XSLT2 will cause a momentary hump due to vendors
trying to find markets for their shiny new tools) in the medium term.
|
