Re: [xml-dev] %x;? - Opportunity for pedantry and exegesis

[Rob Lugt <roblugt@elcel.com>]

Lauren Wood wrote:
>
> I tried the following short document
>
> {!DOCTYPE a SYSTEM "test.dtd"}
> {a}{z}{/z}{y}{/y}{/a}
>
> with the DTD
>
> {!ENTITY % x "z"}
> {!ENTITY % a.content  "(%x;?, y)" }
> {!ELEMENT a %a.content; }
> {!ELEMENT z (#PCDATA)*}
> {!ELEMENT y (#PCDATA)*}
>
> (with curly braces for angle brackets to avoid mailer mangling).
>
> The ElCel validator says the file is valid.
>
> IE 5.5 says it isn't, with the error message:
>
> Invalid character in content model. Line 1, Position 5
>
> ( z ?, y)
> ----^
>
> Is this a known bug? I couldn't find it in the knowledge base on MSDN.
Does this mean
> nobody has looked at XHTML documents with IE 5.5 and tried to validate
them? Or am I
> missing something?
>

Lauren,

I tried this on a machine containing a native IE5.5 and got the same results
as you.  I then tried on another machine with MSXML3 in "replace mode" and
it worked okay, so it seems that this bug has been fixed.  As you could
guess from your example, the bug was a serious one - PEs were being expanded
with an extra space even when included within literals.

However, most people probably haven't installed MSXML3 in replace mode, so
most installed browsers will have this bug.  I'm sure some people must have
tried to validate XHTML documents using IE5, and they probably turned to a
more compliant validator when this exercise failed.  However, AFAIK, this
bug does not prevent the viewing of XHTML documents using IE5.  This is
because XHTML documents are generally served up with a MIME type of
"text/html" which IE5 treats as plain old HTML, quietly ignoring any DOCTYPE
declaration.

~Rob

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