RE: [ubl-ndrsc] Fw: Global vs. local

["Mikkel Hippe Brun" <mhb@schemaworks.com>]

> >>>
> >>> I suggest that the NDR SC considers the following questions:
> >>> 1. Which is more important:
> >>>        a) Easy transformation but inconsistent naming of elements
> right
> >>now,
> >>> and easy transformation (based on type) and inconsistent naming later.
> >>>        B) Difficult transformation and consistent naming right now,
> and
> >>easy
> >>> transformation (based on type) and consistent naming later.
> 
> Good points.  That's one way to look at it, but I presume from
> your discussion that the "easy" and "difficult" adjectives are
> based on using solely XPath technology.

My primary experience is with XSLT transformation. But I believe that type
information will be of value to other methods of transformation. Using the
Post Schema Validation Infoset allows (to my best knowledge) for inspection
of element types. My question above is independent of XPath.
 
> There were a number of trial transformations of XML document sets
> we did involving about a thousand nodes.  Not exceptionally large,
> but the XPath versions took very very long time to do even a simple
> search, although the expressions looked compact.  We're forced
> to re-program the expressions with some other means to take advantage
> of certain boundary conditions that cannot be represented by
> XPath expressions.
>
> The point here is that design phase issues, which are what NDR
> is discussing, can be sensitive to practical considerations,
> but need not be bound by one particular way of doing things,
> if certain recommendations fulfill the "do-the-right-thing"
> objectives.

Your point is taken. Let us ignore XPath and simply consider the ability to
do transformation based on type. I still believe that transformation based
on type (be it XSLT, SAX, DOM or other means) will be used extensively in
the near future. Given that this prediction turns out to be true, the above
questions still applies.

> 
> 
> 
> >>> It may very well be the case, that when UBL "takes off" - XPATH 2.0
> will
> >>be
> >>> a reality and that we may regret a choice of Global element
> declarations.
> 
> We can afford to regret later than to regret now, can we?
> 

We have to live with messages based on our decisions for a long time. If it
turns out in 6 months that we never use our global element declarations for
transformation - we still have to live with the inconsistent naming.

I am still voting for a global declaration approach, but I believe that the
above discussion should be taken into account.

Regards

Mikkel Hippe Brun
SchemaWorks
Ærenprisvej 14, 2820 Gentofte, Denmark
tel.: +45 2081 6923 e-mail: mhb@SchemaWorks.com