|
A
simple example where validation seems like it should always be done: when
an application that depends on its XML conforming to a particular schema
receives XML from an external source. In the real world, we receive XML
purchase orders conforming to the cXML spec over HTTP from a different
organization. Validation quickly gets us started with the processing of
the PO.
In
such a case validation accomplishes the important task of assuring a basic level
of conformance to some specs of the application, and typical applications will
additionally check the message in ways that DTDs or Schemas couldn't necessarily
help with, but that is not to say DTDs or Schemas are
incomplete.
At the
other end of the spectrum would be the pointless tactic of validating the same
file validated 20 seconds ago, the same application validating the same file
sitting on the same server.
If you
are talking about publisher and subscriber applications, the big question is how
close are the two systems to each other? What organization, what network,
what machine, what file? It would range from two distinct applications at
different organizations to two modules of a single application accessing a
single file, and at the one extreme you should validate and the other you
shouldn't.
Max
-----Original Message-----
From: Magick,
Brian [mailto:Brian.Magick@COMPAQ.com]
Sent: Wednesday, October 31,
2001 6:25 PM
To: xml-dev@lists.xml.org
Subject: [xml-dev]
When to Validate XML?
Anyone interested in sharing their
thoughts on under what circumstances you should validate XML?
I think this board has really had
some good discussions lately (i.e. Database Decision Tree, Determining ID-ness
in XML) and this is one issue that still seems to draw the most debate in the
inner-circles of my company.
Some folks feel that validation
should ALWAYS be done regardless of the circumstance (a bit radical it may
seem). Others feel that XML validation is simply a waste of time. Some groups feel that schemas are not
advanced yet to do validation very intelligently (i.e. problems validating
date formats), and that the applications are still the best choice to validate
your data. People involved with
integration feel that if the publisher and the subscriber
applications/databases BOTH do validation that XML validation is an extreme
misuse of resources. And on and on….
With out nit-picking the examples
I just laid out on the table can we share some general thoughts on When to
Validate and When Not to Validate XML?
Who knows, if this is a
successful conversation perhaps we can give Leigh Dodds enough ammo for next
week’s article.
J
Brian
Magick
|