[
Lists Home |
Date Index |
Thread Index
]
Thanks folks. I much appreciate the help here.
The determinism question came up in the context of
a wishlist being sent to the XML Schema group, so
that is there, Michael. By the way, congratulations
on your recent XML Cup award. Very appropriate and
much deserved!
NOTE: If I say schema, I mean Schema or DTD
in what follows.
The multiple schemas question comes up as a *life
among the mammals* issue for standards and specifications
developers.
The question originates in interpreting requirements.
If a schema is normative in a standard or specification,
what exactly does that mean,
or what requirements does that place on the
developer of an implementation of the language?
Even more befuddling for the the developer or
user, what exactly is meant by the 'implementation',
that is, an authoring tool, a browser, a query engine?
The pipeline answer is likely the easiest to present
and understand. The multiple languages answer comes
next, but are part of the same problem when attempting
to create a normative control.
Most don't, AFAIK, usually consider a pipeline when
writing a specification, or if they do, they usually
try to build a one size fits all control which I'll call
the Gorilla Schema (Whatever the Gorilla wants, it gets).
It can result in a control that is hard
to relax or constrain without breaking it. Schema
designers have a lot of freedom and one of these is
to mix runtime information into the language. We
all probably understand the yins and yangs of that
but one of them is to break the scope of the schema
across the pipeline.
The normative schema can become this ponderous gorilla that scares
the hairless apes or so confuses them that they immediately
launch counter efforts to 'simplify', 'make sane',
'reduce to...' and so on that unfortunately turn
political. We are likely familiar with
this event type: the counter-spec. This can become
not just a technical effort, but a movement against
the organization and individuals. As a "His Hers And Ours"
kid, I have an inbuilt reaction against that emotional
response to an uncomfortable outcome of process.
It seems to me that multiple schemas are another best practice
issue related to the discoverable aspects of documents
in use. The question that one might pose is what
relationships these emergent controls should have to the normative
controls. Certainly, one demands that they be compatible
if not compliant (eg, they don't introduce productions that
are invalid in the normative control).
The trick is to separate the emotional reaction against
the organization from the fear of the normative control,
and to get the organization to relax and understand that
multiple schemas don't necessarily mean loss of control.
So we have an education issue.
len
|
