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

["Simon St.Laurent" <simonstl@simonstl.com>]

On 3/25/13 10:00 AM, Toby Considine wrote:
> Do you document the interfaces/messages  that you use internally, even if
> only for yourself?

Yes, of course.  That barely requires schemas, however.

> Do you write down the assumptions you make, and what might be an error, even
> if only for yourself, when you come back in a year?

It depends on the complexity of the problem, though of course you always 
have to allow for the fact that distance from the code will add complexity.

> Is any of this documentation machine readable, to be consumed by your tools?

Probably not.  How many machine readable representations do you actually 
need?  Worse, creating machine readable representations creates the 
temptation to treat that representation as law.

> And when you get hit-by-a-bus (or take another job, or get bored and open a
> goat dairy) what happens? Or if you do not care (because you'll be off with
> the goats...), how do you feel about keeping those systems working when the
> person you rely on when exchanging forbidding stray notes, or answering
> out-of-band queries, or who has said something completely different from the
> vocabulary you've chosen leaves to be a maker of goat cheese...

You're trying to change the subject, presenting a broken culture as a 
sane response to forgetfulness and loss.  Sorry, but that is not an excuse.

> A culture of innovation a culture of compliance, or even a culture of
> straight-jacketing can all exist whether or not we use standard schemas.
> Standard schemas can coexist with ad-hoc schemas. As long as we avoid those
> zealots (and I know them, too) who want to extend their one schema or
> information model to be used in all other spaces, whether they fit or not,
> then the standard specifications add great value. An overbroad
> specification, or a minimal standard badly applied, can certainly reduce
> productivity and creativity. A minimal standard, well used, enhances
> creativity.

Constraints can be wonderful - I'm having a hard time breaking away from 
writing a book because I set the right constraints for it.  It has also 
helped that I changed the constraints to some degree along the way. (And 
that I will continue to change them in response to reader feedback...)

That in no way justifies the "constraints first" model that has 
dominated markup conversations for the last three decades.

> As Ed Crowley observed long ago, "There are seldom good technological
> solutions to behavioral problems". You seem to be focusing on behavioral
> problems, which can be endemic in certain organizations, and on the misuse
> of specifications.

Not quite.  I'm diagnosing XML's role in transmitting those diseases, 
and the impact that role has had on XML's acceptance and usefulness.

Those who won't acknowledge that they are sick rarely respond well to a 
diagnosis.  Those who make their livings from the continuation of the 
illness - well, that's even more complicated, and a key part of the 
problem we've built.

Thanks,
-- 
Simon St.Laurent
http://simonstl.com/