RE: [xml-dev] RE: Concerned about the increasing reliance on XPath

From
Liam R E Quin <>
To
"Costello, Roger L." <>
Date
2011-05-09T21:50:55Z
ID
<>
Thread
RE: [xml-dev] RE: Concerned about the increasing reliance on XPath
On Mon, 2011-05-09 at 15:52 -0400, Costello, Roger L. wrote:
> Hi Mike,
> 
> > What does your "assessment" step really do anyway?  You didn't say.
> 
> There are many ways that one may wish to analyze an XML instance document. For the "purchase XML instance document" we may wish to analyze it to determine:
> 
> - is the value of <Total> a decimal value (and nothing else)
> - will the instance document, when evaluated, trigger unexpected and undesirable actions
> - is the value of <Total> equal to the sum of its Item siblings
> - perform an algebraic data type analysis

At this point you're into what I think of as "pipeline validation",
where you have a series of steps, typically a sequence, and maybe use
XProc to orchestrate it.

For example, one step could calculate Total values, another could go to
a database and populate product descriptions based on part numbers,
another could check the Totals are right :-) and so on...

In most cases there's not much point carrying around precomputed values
- see Third Normal Form for relational databases for why duplicated
information is usually a warning flag.  But computing it on the fly,
processing the augmented document, and then checking the constraints
still hold, that's another matter.

I would almost certainly expect to have the logic _outside_ the document
in most cases, but e.g. an XML-serialized spreadsheet might not.

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Pictures from old books: http://fromoldbooks.org/