Re: [xml-dev] Re: ***SPAM*** [xml-dev] Re: The Goals of XML at 25,and the one thing that XML now needs
> But I suggest that modes preclude many and perhaps most parallel implementation methods,
> which has helped XML adroitly avoid most of the improvements to CPUs in the
> last 25 years. And a technology in that situation is not in a healthy place.
XML and SGML don't have escape characters, owing to their IBM as
opposed to Unix tradition also showing in its commenting syntax
(technically, SGML has MSOCHAR, MSICHAR, MSSCHAR but these are not
assigned in the default concrete syntax). Verbatim inclusion of markup
delimiters is performed through entity references (eg <). So isn't
SGML/XML already doing what you're postulating ie avoiding the need
for arbitrary lookback in assessing whether a given escape isn't
escaped itself etc etc, which would be necessary with backslash
escapes in the Unix way, and which btw can't also be captured by
regexpes, or only in a bounded way?
I think Liam was thinking of the problem of double (or triple) delimiting, such where you want to generate fragments of code in one language (which has, say, lots of & and " and '), using another language (which also uses those delimiters) and then have this in an XML attribute (which also uses those delmiters.) It becomes error prone.
Contrast with some other imaginary approach where, say, you can select the delimiter map for attribute values (like an SGML #USEMAP) and so establish a mode where there is temporarily some less-likely-to-be-found delimiter sequences for " ' and &. This makes the attibute to require be one level less of delimiters.
I agree with Liam that would be a great feature for something (XSLT?). But it is maybe the opposite of what is needed to open up parallel processing, AFAICS.
Or would it be enough to specify an alternative attribute value delimiter of """ which would then allow single " directly in the literal value without using a reference (but still require them for <>&)? ...
Cheers
Rick
