On 12/08/2013 09:13 AM, John Cowan wrote:
> In that case I fall back on the position of making ]]> an error (which
> means, of course, that implementations are free to treat it how they like).

I've changed my mind, and I disagree.  I looked at the specifications for
"XML literals" in XQuery and ECMAScript for XML, and as far as I can tell,
both allow ">" without even mentioning the "]]>" issue.  I also tested
SaxonHE 9.5.3, and it accepts "]]>" without complaint.

If W3C's XQuery allows "]]>" I don't think we should prohibit it, and
I'm not
sure we need to warn about it.

The text in my current draft (in the "Element contents (children)"
section is as follows:

   The characters & and < are special and need to be escaped.

   The character > does not have to be escaped, but it is good style to
always
   do so, as it makes it easier to visually distinguish it from markup.
   (The MicroXML proposal does not even allow unquoted >.)  If an
XML-node value
   containing > in element or attribute content is written, an
implementation
   *should* write the escaped form &gt;.  The XML and HTML 4.x standards
do not
   allow the literal text ]]> in element content, for historical reasons of
   SGML-compatibility. For this reason an implementation of this
specification
   *may* warn if literal ]]> is seen.

I think that says what needs to be said.  I can change the last *may*
to *should* if you prefer.
--
	--Per Bothner
xxxxxx@bothner.com   http://per.bothner.com/