OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: When To Use Schemas (Was RE: infinite depth to namespaces)

"John E. Simpson" wrote:

> :)  In my own defense, what I was really going on about there wasn't
> validation per se, the general usefulness of DTDs/schemata, and so on. It
> was the relentless focus on them -- as though it's impossible to do
> anything at all useful without them. ("The focus on DTDs and XML Schema as
> the hallmark of so-called real XML has done more to damage XML's widespread
> use and popularity [etc.]...") They're extremely useful tools once you've
> got the basics nailed down, but can be a stumbling block en route to that
> point.

I started this paragraph feeling that perhaps that's one of the areas where XML
really is a different beast from SGML and databases, but now I'm not so sure.
My background and instinct lead me to believe that you need to know what "it"
will look like before you start creating it, but that clearly isn't the case
with all XML. Come to think of it, I wrote plenty of SGML DTDs to suit data
sets, so maybe they really aren't that far apart after all.

There's no doubt that there's a real place for the separation between
well-formed and valid. Many people found that conversion to SGML from another
format used to work well when a number of stages were used, but that typically
involved a separate DTD for each stage (or one very loose one). It would have
been very useful to know whether the errors were related to validity or
well-formedness. If they were validity errors, I could adjust my interim DTD -
if they were well-formedness errors, I would probably change my conversion
scripts. That would have added another very useful dimension to the layering
that so obviously made sense in the first place.

> Not wanting to drift off into pedagogical theory or anything, but it's
> generally a question of deductive vs. inductive learning. Teach someone
> well-formed first, and teaching them valid later will be a piece of cake...

That may well be the case overall, though my personal preference is to
understand the rules of the game before I start playing badly. We needn't drift
further I don't think - I can start to hear the waves crashing on the rocks

> Mostly I just wanted a question that would trigger some righteous
> indignation in me... even if the hypothetical newbie didn't get her actual
> question answered in the process!

I feel that she should have been happy with the answer overall. If one message
from your article is that the decision of whether to validate should be driven
by the current requirements rather than necessarily by the data, I definately


Marcus Carr                      email:  mrc@allette.com.au
Allette Systems (Australia)      www:    http://www.allette.com.au
"Everything should be made as simple as possible, but not simpler."
       - Einstein