[
Lists Home |
Date Index |
Thread Index
]
- From: Dan Mabbutt <Seigfried@msn.com>
- To: ",XML Developers List" <xml-dev@XML.ORG>
- Date: Thu, 13 Jul 2000 14:28:25 -0600
Thanks to Rick Jelliffe for a well written explanation of one-half of my
confusion. And thanks to Fabio Arciniegas for describing the other half.
My conclusion so far is that, yes, there are two groups using the same name
for what they're doing - even though what they're doing is quite different.
From my XML novice point of view, I think that the folks in the "design
pattern" group ought to think about changing their terminology. Even some
of their references don't use it, and it is at least initially confusing.
Simon St. Laurent started a thread a while ago about, "power uses of XML vs.
simple uses of XML." Although it degenerated into a dispute about the value
of namespaces, the early part of the thread was a very illuminating
discussion about the possibility that standards were proceeding beyond the
needs of developers and business, and the split personality of W3C in both
standardizing known and proven practice, while at the same time pushing the
frontiers of unproven theory. That is the kind of soul-searching
introspection that is necessary to keep an organization like the W3C from
sinking into a morass of irrelevance, or worse - obstruction. Taking a stand
on mundane day-to-day matters like this ... although not likely to win any
prizes for technical brilliance ... is an example of where an organization
like the W3C can actually contribute.
-----Original Message-----
From: ricko@mail.geotempo.com [mailto:ricko@mail.geotempo.com]On Behalf Of
Rick JELLIFFE
Sent: Thursday, July 13, 2000 1:14 PM
To: ,XML Developers List
Subject: Re: Patterns - What is it?
Dan Mabbutt wrote:
> The first group is represented by the page http://www.xmlpatterns.com/. A
> brief review of this page and some of the references on it suggest that a
> "Pattern" is sort of a "best practice" or "archetype software algorithm."
An "design pattern" is ultimately a literary form, used to capture
experience by forcing certain answers to certain questions about a topic
to be provided. The form is loose.
Design patterns were developed for architecture, but they have had a
major impact on object-oriented programming. When you read someone
writing "this is a visitor class" or "that is a factory method" they are
refering to a well-known design pattern.
The reason design patterns are important is that without establishing a
common vocabulary there really can be no XML "professionals". Without
such a vocabulary, people will attempt to use a secondary vocabulary,
which will usually be OO or databases or HTML; those vocabularies may
include enough baggage that they may allow communication while
preventing understanding.
For example, one of the most basic patterns is the one variously called
the "document skeleton" or "document shell": that is the simple
structure that whenever information units grow to a certain level of
size or have a strong semantic or processing cohesiveness compared to
the outside material, we will see a common structure
thing
- head (with metadata)
- body (with data)
At the lowest level, this is actually supported in the element tag
syntax, but it scales all the way up so that documents will often be
structured at the highest level in a similar way. I think Fabio has a
design pattern for this at www.xmlpatterns.com
It is such a common pattern, but I doubt that if we asked people on
XML-DEV or any of the W3C working groups what that was called, that we
would get a recognised term. Without such a controlled vocabulary,
discussions are forced to start from the beginning with actual examples:
this is such a burdon that it prevents discussion of issues. This will
only get worse as more people con on board who cannot write content
models: not only will it be difficult to write prose, but difficult to
write formal notations.
Rick Jelliffe
|