Re: [xml-dev] Not using mixed content? Then don't use XML

[Michael Sokolov <msokolov@safaribooksonline.com>]

On 3/25/13 9:06 PM, Simon St.Laurent wrote:
> (Rarer even than the schema-free respondents to this and related
> threads, whose presence I've been excited to discover.)
>
>
Here's another: in the last 10-15 years of working w/XML (mostly from 
publishers), we've almost entirely avoided schema, and although many 
(most) of the documents we've received have had DTDs associated, when 
possible we've ignored them.  The only time we absolutely had to learn 
about schema was when dealing with SOAP.  We have schema-lovers in our 
office, too: they tend to be the people looking for someone to blame 
when things go wrong - they want to have some way of knowing whose fault 
is it when something fails. That's valid, but isn't really a helpful 
perspective for a creative problem-solver.

I've often looked at schemas and DTDs as a potentially useful guide to a 
new data format (documentation), and to some extent they can be. But we 
often find that they fail in two ways:

On the one hand the industry-standard, widely-used, schemas have usually 
been defined so broadly that they encompass a huge amount of possible 
markup that may never be encountered in (our) practice. This makes them 
useless as constraints on development since the cost of "support" for an 
entire schema is never really warranted. So we can't really leverage 
existing schemas easily, since we still end up having to analyze the 
actual corpus in order to understand which subset of the markup is used 
in practice, and this varies wildly from customer to customer.

And on the other hand, schemas (and certainly DTDs), for all of their 
expressive power, often fail to capture the variety of interpretations 
given to markup.  I'm reminded of the customer that really wanted to put 
call-outs containing primary source document fragments inline in a 
document, but didn't have any appropriate tag defined in the DTD for 
that purpose.  Rather than simply being free to invent one, they choose 
to abuse the footnote tag for the purpose and added some completely 
unrelated convention to distinguish these structures from traditional 
footnotes.  Having become accustomed to this convention, they went on to 
invent other uses for footnotes as well - marketing blurbs went in 
there, too IIRC.  I'm sure markup abuse like this goes on all the time, 
and I think the slavish adherence to standards is partly to blame.

-Mike Sokolov