John Cowan wrote:

>  Jonathan Borden wrote:
> 
>  > But it is not hard to describe, that is my point. The XML 1.0 production
>  > rules create a parse tree which exactly describes the source document,
down
>  > to the byte. This is what I would call the XML property set, and from
this
>  > one can subset to one's desire.
> 
>  Okay, I understand now.  Actually the production-rule level is *not* the
>  bottom parsing level, because for the most part PE-references have already
>  been removed.

I new there was a reason why you got this job :-)

PE-Reference is production [69] so it can be represented in a production
rule tree, the issue is that the elementDecl tree cannot be built until the
PE-Reference is expanded.

I would handle this as an <alt>  with both the PE-Reference and following
character content, and its expanded form side by side.

e.g.

<PEDecl p="72"> 
    <seq> 
        <S> <c char=" "/></S>
        <Name> element</Name>
        <S> <c char=" "/></S>
        <EntityDef> <string value="<!ELEMENT" /></EntityDef>
     </seq> 
</PEDecl> 

and

<alt> 
 <seq> 
   <PEReference> <Name>element</Name></PEReference>
   <S char=" "/> <string value="FOO EMPTY>" />
  </seq> 
 <elementDecl> 
    <seq> 
        <S/> 
        <Name> FOO</Name>
        <S/> 
        <contentSpec> EMPTY</contentSpec>
    </seq> 
 </elementDecl> 
</alt> 


>  Internal PEs are not really structural in XML;
>  it is a mere validity constraint that requires DTD constructs to begin
>  and end in the same PE, and
> 
>  <!ENTITY % element "<!ELEMENT">
>  %element; FOO EMPTY>
> 
>  is a well-formed though not valid external subset.
> 
>  > Doesn't it make the most sense to subset from the full description,
rather
>  > than both add to and subtract from a partial description?
> 
>  The question then remains, just what is
>  useful structure and what is not?  I have done my best to answer that
>  question.

Useful for the DOM and XPath, I agree, but since these already exist, and
define the middle subset, now is the time to define the full set.

I have no problem with what the Infoset is trying to accomplish as a subset
of XML 1.0 + Names (just as I have no problem with what Common XML etc are
trying to accomplish), but this activity needs to come from a complete base
down to X,Y or Z subset each of which are useful.

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org