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

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

On 4/10/13 8:38 AM, James Fuller wrote:
>> The disease is the brittle legalism that schemas encourage and then enforce.
>
> right, but as most know on this list ... schemas are optional ... I
> dont understand your rigid conflation with schema usage and XML.
> Personally I think schema application is just a matter of timing ...
> thats nice thing about XML stack eg you dont have to use a schema and
> you can if you want too enshrine things once volatility in the data
> model calms down.

I think you've mistaken expert usage for the popular reality. 
Programmers may get to make decisions about when and where they apply 
schemas, but unfortunately schemas dominate the conversations about how 
to use XML in the first place.

> So to sum up and put simply and to encapsulate knowledge;
>
>       You think the usage of schemas are a bad idea with XML, their
> usage is so widespread to be considered a terrible disease and we need
> to be purged of their usage ?

Yes.  There may be corner cases where they are actually beneficial, but 
the risks of descriptive markup turning into prescriptive markup are 
substantial.

> I could just as easily say you are laying down the tenets for some
> form of Spanish Inquisition,

You could, but you would be wrong.  The Spanish Inquisition had power. 
I am merely a heretic.

> but hell I agree with you on most points
> about usage of schemas ... you may even dismiss them as a form of
> early optimisation.

Yes, that's a gentle way of dismissing them, but worthwhile.

>> We mistook this disease for the true promise of XML - agreements!
>> transparency! - and encouraged it to rage in epidemic form, driving
>> developers and organizations away from XML and markup entirely.
>
> in the xml circles I am/was running, terms such as 'true promise'
> never popped up as a concept.

What XML circles are/were those?  Again, I'm happy to find exceptions - 
they're just rarely as exceptional as people seem to believe when their 
work is being criticized.

>> It is not the only issue driving developers away from the XML ecosystem, as
>> previously discussed.  However, it seems to be the issue made invisible by
>> our blinders, the piece we just can't imagine is really a problem.
>
> once again ... the statement 'driving developers away' makes me think
> you value adoption over all things... there are corners of computer
> science which are fascinating, correct and fully under adopted.

I absolutely value adoption.  It isn't everything - I just wrote a book 
on the strange wondrous corner that is Erlang.  It is growing, but I 
won't be saddened if its future is more to teach than to deploy.

Beautiful is good.  Beautiful hidden away in a dark corner where few can 
see the beauty, buried under layers of dross... well, I'm sure there's a 
fine story waiting to be written there.

> however, XML is not one of them ... XML has huge adoption, we are just
> not at the very top table anymore and really probably never should
> have been there in the first place.

XML is dwindling.

Yes, the bubble in its use was a perverse side-effect of opening up 
possibilities that had been closed for too long.

However, it is fading far faster than its actual qualities suggest (to 
me) that it should.  Markup offers a unique combination of structure and 
flexibility, and the latter is a story we have barely begun to tell.

I suppose that burdening JSON with schemas could be an excellent way to 
cripple it and make people think XML wasn't so bad.  However, except for 
minor verbosity issues, there is no reason at all that markup couldn't 
have had JSON's place in the development ecosystem.  If only we hadn't 
tripped over our own feet...

A key part of my longer-term argument, though, is that markup can 
actually do a better job of supporting digital conversation than JSON 
(etc.) does today.

That requires a rethink of how we process markup and develop 
vocabularies, however.

And that requires throwing schemas over the side.

> To make a football analogy ... we
> will be in the 1st division though for the next 30 years.

It does a university little good to spend a fortune on a football 
stadium for a team whose record gets worse every year.  I don't see XML 
surviving beyond legacy work for more than a decade on its current course.

> doesn't sound complicated to me ... just dont use schemas ? oh but
> then you will start using code to constrain your data ... after a few
> projects eventually you will build up abstractions, flexible
> mechanisms and code libraries that start looking like what schemas do
> already ... then you will ask yourself perhaps its better to adopt
> something thats been discussed generally as a standard .... ad
> infinitum.

Sorry.  It's clear that I have to describe a path forward in more 
detail.  If you think schemas are inevitable, I clearly have a lot more 
work to do.

> Schemas are only as powerful as the programming language (and idioms)
> and hardware architectures they are developed upon ... I get the
> feeling (and concur with you) we need more expressive approaches ... I
> am sure there has been a lot of work and research in academic
> literature pointing how hardware must change (which is the trigger
> really for any 'real' change in our industry).
>
> I actually think what you want is new hardware to enable this revolution.

Not remotely.  It's up to the people, not the hardware.  The flaw is not 
in the machines, but in ourselves.

(The machines have their own flaws, but that is a different conversation.)

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