This is *excellent*! Can I quote you, or will you make it publicly
available?
Andrew
109b SE 4th Av
Gainesville
FL 32601
Cell: 352-870-6661
http://www.andrewmaben.com
[EMAIL PROTECTED]
"In a well designed user interface, the user should not need
instructions."
On Feb 9, 2007, at 3:34 AM, Breton Slivka wrote:
A schema is only part of the story. A schema can only help
determine whether a given file is valid. It does not contain any
instructions about what the data means. In fact, I don't believe
that anyone has discovered a way to really teach a computer what
any information means. It's all just programmers defining various
things to do with the data.
This is somewhat off the point though. Once we establish that
meaning is something that only exists inside the human mind, and in
the practical case, the mind of a particular software author, the
problem becomes that of reliably deciphering the *same* meaning
from the same body of data. So the goal of any reusable data format
should be to reliably communicate this meaning to potential
software authors. It is my point that this cannot reasonably be
done with a set of XML tags alone. Nor would a schema do much good
towards this end.
To accomplish the goal of communicating meaning, one needs a body
of Natural Language, in this case, English, with its words in the
context of its native Grammar and Syntax. This body should clearly
explain the purpose of the format in question, the meaning of each
of its elements, an define procedures for dealing with these
elements. It needs to be in a Natural language since human
programmers are its target parsers. This body of Natural language
is what I refer to as a "Spec", which is an abbreviation for
"Specification. In that it specifies the meaning and nature of a
particular format.
However I believe it is not good enough for individual format
authors to simply invent a format, and explain its meaning. The
goal is not simply isolated storage and retrieval in the case of
the internet, but reliable interchange. If we each individually
author our own slightly different formats with overlapping goals,
it becomes a problem to coherently exchange data between these
various systems designed for divergent definitions of data. This is
why standards are important. Not because they feel good, but
because agreeing on specific formats and conforming to a single
stated meaning for that format is the *only* path towards reliable
data interchange. Everything else is just a wank.
*******************************************************************
List Guidelines: http://webstandardsgroup.org/mail/guidelines.cfm
Unsubscribe: http://webstandardsgroup.org/join/unsubscribe.cfm
Help: [EMAIL PROTECTED]
*******************************************************************
