Re: modularity in specifications

At 01:55 PM 3/18/02 -0500, Karl Dubost wrote:
>At 12:04 -0500 2002-03-18, Lynne Rosenthal wrote:
>>(1)  what does modularity mean for a specification  - (this is something
>Several issues here.

This is probably accurate -- I suspect that we will see entries into the 
Issues List, as Spec Guidelines develops and starts getting review 
attention.  (Actually, we already have a long-dormant issue #13 [1], about 
inter-standard or multi-standard interoperability, which touches on the 
subject.  But from an interoperability perspective, as opposed to a Spec 
Guidelines perspective).

One further comment in-line...

>- Modularity as different pieces of a spec:
>         For example, XHTML Modularization, we are conformant to the Text 
> Module but not the table one, ... So what does it mean to be conformant, 
> it's why I think the specification must declare the way they want people 
> claim their conformance, like UAAG did.
>- Modularity as different features of a spec:
>         Imagine a "Braille" implementation of SVG, you have a pin-point 
> board which draws in relief a square, a triangle, etc, so which draws a 
> 3D shape of a 2D vector graphics. It has no sense to implement colors in 
> this kind of things. So what the conformance means in this case and how 
> you can claim it.
>- Modularity as a Composite document:
>         What's a composite document, A document done with namespaces, do 
> you need always a schema or a DTD to validate this kind of documents. If 
> namespace are sufficient what does that mean for a player to be 
> conformant or not. For example a player that will not accept and strip 
> every tags it doesn't understand and so destroy the integrity of the 
> document in memory, is it a conformant player.

SVG, for example, explicitly recognizes in its conformance clause [2] that 
SVG content may be part of a composite document.  However, it only 
addresses conformance of the SVG fragments themselves -- rightly so.  It 
would seem logically to be the responsibility of the "composite document 
standard" to define test assertions that could lead, for example, to a test 
suite for a player of the whole document.  Only problem:  what is the 
"composite document standard"?  We could have XHTML documents with embedded 
SVG fragments, or we could have SVG documents with XHTML fragments.  SVG 
1.0 carefully defines what is correct handling of the SVG fragments in both 
cases, but it cannot make any detailed testable assertions about the XHTML 
fragments.  All it can say is "...foreign object, to be interpreted by 
appropriate handler, in this spatial area of the rendered SVG document", 
for example.



Received on Tuesday, 19 March 2002 17:03:08 UTC