Quoting Laurens Holst <lholst@students.cs.uu.nl>:
>> 7. If it encounters an entity reference (other than one of the 
>> entities defined in this recommendation or in the XML 
>> recommendation) for which the user agent has processed no 
>> declaration (which could happen if the declaration is in the 
>> external subset which the user agent hasn't read), the entity 
>> reference should be processed as the characters (starting with the 
>> ampersand and ending with the semi-colon) that make up the entity 
>> reference.
>
> Correct me if Iâ??m wrong, but doesnâ??t the XML specification define 
> different processing rules (i.e., throw an error)? This would be 
> orthogonal to item 1 of that same section, and a bad idea, too.

No, that's just what one UAs has implemented as behavior for undeclared
entities, but not what should happen when you encounter them.


> In section 3.2, list item 3 of the XHTML specification, it says the 
> following:
>> 3. When a user agent processes an XHTML document as generic XML, it 
>> shall only recognize attributes of type |ID| (i.e. the |id| 
>> attribute on most XHTML elements) as fragment identifiers.
>
> What about xml:id, then? Why this restriction?

I don't think the part between parenthesis really imposes a restriction. It is
never said that other attributes can't do the same thing.


-- 
Anne van Kesteren
<http://annevankesteren.nl/ >