Re: [xml-dev] hackable xml

[Andrew Welch <andrew.j.welch@gmail.com>]

On 26 July 2010 12:19, David Carlisle <davidc@nag.co.uk> wrote:
> On 26/07/2010 12:08, Andrew Welch wrote:
>>
>> 5. Lone inbuilt entites such as "&" in the lexical XML are
>> automatically parsed as&amp; and not an error (#2 above might enable
>> this). Same goes for a lone "<".
>
> why this? With it it means that the resulting files are not XML, so the new
> format couldn't use the various xml processing tools unles sthey allow a
> custom parser to be provided, in which case you don't need a new format, you
> could presumably hook up a json 9or html5) parser that reported xml sax
> events.

I thought about that, and it was dependent on the crucial namespace
decision, which if it was the "significant prefix" option would break
compatibility with existing parsers anyway.  I did group 1 to 4 as
implementable now using a "style guide", and 5 and 6 as requiring new
parsers/serialisers, but then removed it to keep it simple... oh well.

For this one, the goal is to avoid a lone "& " breaking the parse, as
that's a common problem which may well be avoidable if non-inbuilt
entity refs are dopped... I don't know if that possible or not, its
not that significant.


> Without it, it's a variant in the usual simplifications going back to sml
> (if not before) the usual reason that nothing actually happens is that while
> it's easy to agree that a simplification would be good, it's a lot harder to
> agree on the details 9as this message shows:-)

Googling "SML" shows it was over 10 years ago!

http://www.xml.com/pub/a/1999/11/sml/index.html

It does seem overly concerned about attributes for some reason,
probably because namespaces hadn't been invented at that point?

Either way, 10 years down the line the shine has worn off XML and
opinions will have changed.  There is a definite need for this, and
everyone's lives would be a little easier if it happened.


-- 
Andrew Welch
http://andrewjwelch.com
Kernow: http://kernowforsaxon.sf.net/