[
Lists Home |
Date Index |
Thread Index
]
At 12:56 PM +0100 6/6/04, Dave Pawson wrote:
>What processing expectation should we have for such 'extensions' Elliotte?
> Its foreign to an expected schema, not previously met... What is
>your definition of robust and flexible? What one might call robust
>and flexible, others might call guesswork?
This response demonstrates yet another common fallacy in software
design. There are unexamined principles at the foundation of your
question which are so deeply ingrained in your thinking that it
doesn't occur to you that they need to be examined or justified, but
they do.
The fallacy here is that a document has some sort of processing
expectation, but this is simply not true in the heterogeneous world
of the Internet. The document is what it is, and will be processed
differently by different actors. I likely do not want to do the same
thing with the same document as you do, nor is it necessary that I do
so. The demand that we provide and adhere to schemas is often little
more than a demand that we process documents in only certain
preapproved ways. That is a fundamentally limited perspective. It may
work within one program or a small organization. However, it does not
scale to the needs of large organizations and groups of organizations
with different, unique objectives.
You are assuming that the extensions must be processed because
they're there. I disagree. If I don't need them, I am free to ignore
them. My only concern is whether the document contains what I need in
order to perform my task. You likely have different requirements for
that document than I do. I do not guess how to handle anything. I
take what I need, and ignore the rest.
>>Sometimes the answer, is "I don't know" and the document may need
>>to be kicked to a human for further analysis.
>
>Which some might equate to 'fall over and die'?
Absolutely not. The fact is computers aren't that smart, and robust
systems allow and prepare for human intervention. In practice, most
debugged and deployed systems rarely require human intervention of
this sort. However, when they do (and sooner or later they all do) it
is better to be ready for it and acknowledge it rather than silently
drop the problem on the floor by rejecting invalid documents and
claiming it's not your problem.
>I think the SGML world got it right on this one.
As proven by the massive success of SGML, and the complete failure of XML. :-)
--
Elliotte Rusty Harold
elharo@metalab.unc.edu
Effective XML (Addison-Wesley, 2003)
http://www.cafeconleche.org/books/effectivexml
http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA
|