[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: XML=WAP? And DOA?

• To: xml-dev@l...
• Subject: RE: XML=WAP? And DOA?
• From: "Atchley, John" <John.Atchley@f...>
• Date: Mon, 14 Jan 2002 10:56:03 -0500

Okay, I'm fairly new to the XML game but here's my two-cents worth...

Re. the examples below: XML seems better equipped to handle hierarchical
data than CSV. I.e.:

<Something>
	<Date>...</Date>
</Something>
<AnotherThing>
	<Person>
		<Name>...<Surname>...</Surname></Name>
		<Residence>
			<Date>...</Date>
			<Location>
				<Street>...</Street>
				<City>...</City>
				<State>...</State>
			</Location>
		</Residence>
		<Residence>
			<Date>...</Date>
			<Location>
				<Street>...</Street>
				<City>...</City>
				<State>...</State>
			</Location>
		</Residence>
	</Person>
</AnotherThing>

The above, where various elements might appear zero or more times, would be
a bit clumsy in CSV.
 
That said, I think "extensibility" is a more important attribute than
"self-describing."  I see flexibility as XML's greatest strength -
ironically it seems that many of the "advances" in related fields serve
mostly to constrain or destroy that flexibility by burdening XML with
validation info that more rightfully should be in the application domain.

I thought XML was supposed to be a data markup language - an extensible
transport mechanism, if you will.  It seems everybody wants to make it
something it's not, i.e. a complete application development framework.  

It strikes me that cases where data format is closely constrained are rather
poor applications for XML.  Maybe I'm wrong, but it seems to me that if you
can validate a complete XML document against some existing schema then XML
may not have been the best transport mechanism!  If you are developing a
system for sales reps to submit orders to the home office, for example, why
in the world would you use XML schemas?  Both ends of the data connection
(client and server, if you will) have to know all the details about the data
model - and both ends have to be modified if the data model changes, so
there seems to be little to gain by torturing an extensible data model into
a rigid one.  Yes, if you are going to take this example to a b-to-b
environment you need to publish a data model, but the question still remains
"why in XML?"

In my mind, XML is most useful in those cases where you need to store and
exchange information of a hierarchical nature where the content is not fully
known in advance.  Consider for example an application that must receive,
store, and integrate data of varying types from a wide variety of sources -
such an application can be written to store everything it receives complete
in it's hierarchical marked-up form and to "pick through" such data to
identify those elements that it understands and prepare reports as needed.
It may be useful, perhaps even necessary, to define a minimal set of
primitive tags to assist applications in coordinating activities, but as
soon as you rigidly enforce conformance with a specific predefined data
model it seems that you've just discarded XML's greatest, and perhaps only
significant, strength - flexibility.

A document abstraction repository might be a good example of an application
where XML can shine.  Such a service might receive both abstracts and
queries from a wide variety of client applications.  A small set of
"primitive" tags might be defined for convenience (<Location> <Person>
<Date> <Industry> and so on) but you would by no means limit incoming
abstracts to using only the predefined tags.  Predefined tags would be more
in the nature of "hints" so to speak and _might_ be pre-indexed by the
repository for faster search.

This latter application could be implemented using a CSV transport, but the
repository itself would have to use some hierarchical mechanism similar to
XML to store the richly marked-up data, so one might as well use XML for the
transport as well.

Our abstraction repository could be written with enough "intelligence" to
dynamically alter its behaviour based on activity.  For example, periodic
analysis of the stored data might indicate that the (non-predefined) tag
"<Aviation>" has been used a lot so the repository would pre-index that tag
for faster queries, etc.

Finally, if XML does go DOA, I think the cause will be a backlash from XML
projects failing because XML was a poor choice for those particular projects
to begin with.

John Atchley
Senior Software Analyst
Engineering and Design, Courseware Support
FlightSafety International, Inc.
mailto:John.Atchley@F...


-----Original Message-----
From: Jens Jakob Andersen, PDI [mailto:jens.jakob.andersen@p...]
Sent: Monday, January 14, 2002 7:51 AM
To: Al Snell
Cc: The Deviants
Subject: SV: SV: SV:  XML=WAP? And DOA?


Hi Al

> > This is one of the hypes about XML, that I'd like to defuse. XML is
> > not any more self-describing than CSV files. E.g.
> >
> > <99874987kjhk>
> > 	<gnygngyasdada>
> > 		What is this?
> > 	</gnygngyasdada>
> > </99874987kjhk>
> 
> Likewise:
> 
> "Name","Email","Web"
> "Alaric Snell","alaric@a...","http://www.alaric-snell.com"
> 
Really good point.

JJ

-----------------------------------------------------------------
The xml-dev list is sponsored by XML.org <http://www.xml.org>, an
initiative of OASIS <http://www.oasis-open.org>

The list archives are at http://lists.xml.org/archives/xml-dev/

To subscribe or unsubscribe from this list use the subscription
manager: <http://lists.xml.org/ob/adm.pl>

• Follow-Ups:
  • RE: XML=WAP? And DOA?
    • From: Al Snell <alaric@a...>

• Prev by Date: RE: An Architecture for Limericks (was Limericks, Stupidity, and Reality)
• Next by Date: RE: XML=WAP? And DOA?
• Previous by thread: RE: XML=WAP? And DOA?
• Next by thread: RE: XML=WAP? And DOA?
• Index(es):
  • Date
  • Thread