Re: Response to LC-61 Issue]

Martin Bryan writes:

>> If measurements are not included as valid XML 
>> datatypes they will need to be defined as such 
>> by other organizations. Each such organization 
>> will invent its own mechanism for this and 
>> these will not interwork. 

With respect, I disagree.  The fact that we have not undertaken the 
difficult job of standardizing a richer set of measurement types does not 
preclude concerned individuals and organizations from doing so.  In fact, 
we have gone to great lengths to provide the mechanisms one would need. 
For example, we give you the framework for sharing not just string types 
such as:

        <--dimensioned extent as simple type -->
        <length>5 inches</length>


but also:

        <--dimensioned extent as complex type -->
        <length>
                <value>5</value>
                <units>inches</units>
        </length>

which many XML users would consider to be more robust markup. 

In both cases you can declare a type, associate it with a namespace, and 
promote use of that namespace and type as widely as you would like.  I 
think I speak for at least most individuals in our workgroup in saying 
that we very much hope that in a broad variety of disciplines, the 
appropriate standardization will take place, and that where sensible even 
diverse disciplines will cooperate to develop shared types of mutual 
concern.  On the other hand, doing this right can be extremely difficult. 
As we all know, the measurements of length that a physicist and a 
typesetter considered to be fundamental and important are strongly 
interrelated, but in the end quite different (picas vs. angstroms?).  Yes, 
we could have attempted to make all of these fundamental predefined types 
for schemas.  Ignoring the considerable design and discussion time 
required, the result would considerably grow the size and implementation 
complexity, as well as the testing requirements for all implementers and 
many users of XML schemas.

We therefore have tried to draw the line at types such as integer and 
string which have extremely broad applicability.  Certainly, there is not 
unanimity in the workgroup that we have drawn the line at the right place, 
but speaking for myself I think we are close.

(BTW: I hope it's clear that in both of the examples above, the element 
name <length> is just for illustration.  A whole point of having either a 
simple or complex type is that you could apply equally well to <width>, 
etc.)  Thank you very much.

------------------------------------------------------------------------
Noah Mendelsohn                                    Voice: 1-617-693-4036
Lotus Development Corp.                            Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------------

Received on Friday, 29 September 2000 10:29:11 UTC