[XPS-devel] XPL Schema Complexity Reduction

[Reid S. <re...@re...>]

*This message was transferred with a trial version of CommuniGate(tm) Pro*
XPSers,

I've done a lot of thinking about the XPL Schema, as published in
release 0.1.1, and have made some decisions to reduce its complexity.
However, before just charging off and doing it, I thought I'd run my
thinking by you for your opinions. Please respond as quickly as possible
as I'm trying to get 0.2 out in a few weeks and this schema change needs
to be included.

After I published 0.1.1, Victor Kirk brought up a very good point: that
there is no need to support an abstract schema from which multiple
natural language schemas can be derived because (a) it thwarts software
reuse if everyone is developing code in different natural languages and
(b) multiple natural languages can still be supported via XSLT
transformations.

One of my goals for XPL has always been to make it easier to program for
non-English speaking people.  However, Vic's two points keep ringing in
my ears. By making the core XPL language based on English only we can
reduce complexity (and thus speed up translation) of XPL significantly.
It will also make writing the XPL compiler easier. And, in doing this,
not much is lost if we can still create natural language variants of the
XPL core for those who wish to program using language keywords (element
names, really) that use terms in their native language.

So, the decision I've made is to modify the schema as follows:

        1. Ensure there are no attributes in the language as it is more
           difficult to create derivative alternate language schemas
           since XML schema doesn't support extension of attributes
           (this is a shortcoming of XML Schema that needs to be fixed!)
        2. Eliminate all the abstract elements (everything ending in
           .elem) and replace them with their English language
           counterparts.
        3. Keep all the complex/simple type definitions but define them
           in terms of concrete English language elements instead of the
           abstract names.
        4. Combine XPL-Abstract.xsd and XPL-en.xsd into one file,
           XPL.xsd. 

I estimate that these changes will eliminate about 1/3 of the schema
definitions and make it much easier to grasp to the novice.

So, I ask you to respond to the following questions:

        1. Does anyone feel strongly about keeping the abstract schema?
        2. Is anyone dead set against using English as the natural
           language for core XPL?
        3. How much demand for programming in languages that use
           non-English keywords would you expect?
        4. Do you see any fatal flaws in what I'm proposing above?

Thanks in advance for your timely responses.

Best Regards,

Reid