Re: Is whitespace within general entities ignorable?

[Rob Lugt <roblugt@elcel.com>]

Christopher R. Maden wrote:-

> ><?xml version = "1.0"?>
> ><!DOCTYPE test [
> ><!ELEMENT test (child)*>
> ><!ENTITY entws "   ">
> >]>
> ><test>
> >  &entws;
> ></test>
>
> I believe your example is legal, to contradict John Cowan.  ("Ask two
> rabbis, get three answers.")  He mentions that a CDATA marked section
> containing only white space is not legal, and that's explicitly true per
> VC: Element Valid, item number 2.  However, that section does not say that
> entities containing whitespace don't count.
>
> See, however, clause 4.4, which indicates that your reference is
> "Included", and 4.4.2, which indicates that an Included entity "is
> retrieved and processed, in place of the reference itself, as though it
> were part of the document at the location the reference was recognized."
I
> believe that your entity reference is recognized, its whitespace
> substituted as though it had occurred there, and then the content model
> evaluated per VC: Element Valid.  It passes.

Chris, I think this is a strong argument and I, for one, am convinced.  This
also seems to broadly agree with the way that existing implementations have
dealt with the issue.  However, if we loyally stick to the logic from clause
4.4 we can extend the discussion to character references (which are also
'included').  In that case we may be presented with this situation:-

<?xml version = "1.0"?>
<!DOCTYPE tests [
<!ELEMENT tests (test)*>
<!ELEMENT test (child)*>
]>
<tests>
  <test>&#32;</test>  <!-- valid -->
  <test>&#33;</test>  <!-- invalid -->
</tests>

Do you agree that a validating processor should allow the first case and
disallow the second?  Does this raise any eyebrows with the rest of the
list?

Regards
Rob Lugt
ElCel Technology
http://www.elcel.com