[
Lists Home |
Date Index |
Thread Index
]
The past week or so of discussion has left me asking even more questions.
I'm starting to wonder if perhaps we have gotten a little too carried way
with the notion of "declarative approach good, processing description
evil". To some extent, identifying what something is can be more portable
than specifying in detail how it fits into a process. Unfortunately, I
think we've confused the labeling process with an almost mystical notion of
"we bless this X as a Y, so therefore it is a Y." (Or have I just spent too
much time with Perl programmers lately?)
There are some very blurry lines in all of this, and I suspect everyone
sees the lines differently. Developers who feel they have more control
over their information, or who are more comfortable with the kinds of
control provided by W3C XML Schema, seem more likely to comfortable with
"we bless this X as a Y, so therefore it is a Y." Developers with less
control over their information, and those for whom the WXS types aren't a
natural fit, seem more likely to be thinking about a process for moving
from labeled content to value spaces which may or may not be particular to
given applications.
I'm still thinking about approaches to sharing understandings of processing
(in my case, two-directional relations between lexical representations and
value spaces), but it seems that much of the community could use a set of
tools which goes well beyond "we bless this X as a Y, so therefore it is
(must be) a Y." Regular Fragmentations was a step in that direction, but I
think there's a lot more to think about. I suspect that some combination
of regular expressions (perhaps extended by the range algebra Gavin Thomas
Nicol presented at Extreme on Tuesday) and RELAX NG as a framework for
constraints may be a useful way forward.
Finding 80/20 here seems critical. I don't believe that it's possible to
reach a general 80/20 point using the disconnected set of primitives that
W3C XML Schema Part 2 uses as its foundation, and that may be part of why
WXS elicits so many strong reactions. At the same time, I'm not sure
there's any good general way to address two of the tougher cases we've
heard already - prime numbers and irrational numbers.
(In relation to a prior message I posted, I don't think that "sharing"
necessarily means "binding contracts". I suspect there's also an 80/20
point that can be reached by letting developers share local understandings
of processing and criteria without expecting or requiring them to use it as
a data-binding straitjacket.)
Simon St.Laurent
"Every day in every way I'm getting better and better." - Emile Coue
|
