Re: XML5: Re: [xml-dev] MicroXML

[David Carlisle <davidc@nag.co.uk>]

On 14/12/2010 01:29, rjelliffe wrote:
> Is Henri suggesting (and I don't want to say it is a bad idea...) that
> really there already is a subset of XML implemented and in the field:
> that subset of XML that HTML5 parsers will support? I think Henri wrote
> "XML5" at one stage as a typo, but it fits.

"xml5" wasn't a typo, it was a reference to

http://code.google.com/p/xml5/

The trouble is that HTML5 parsers will parse a large subset of XML but 
produce radically different parse trees, leading to silent data corruption.

It's virtually impossible for a human to figure out what parse tree an 
html5 parser will produce from a fragment of well formed XML, you just 
have to suck it and see.

Some random examples

1)

<p><a id="foo"/> <b>xx</b></p>

parses as

<p><a id="foo"> <b>xx</b></a></p><a id="foo">
</a>


2)

<p><math><mfrac><mn>1</mn><span>x</span></mfrac></math></p>

parses as

<p><math><mfrac><mn>1</mn></mfrac></math><span>x</span></p>



3)

<p><maths><mfrac><mn>1</mn><span>x</span></mfrac></maths></p>

which has the same structure as (2), but  a different element name, 
parses as

<p><maths><mfrac><mn>1</mn><span>x</span></mfrac></maths></p>



The polyglot spec (which still has its own problems) tries to lead 
people away from these traps, but documenting that they exist doesn't 
really help unless all the tools are updated. the costs of that compared 
to the costs of html specifying a sane parsing mode in standards mode 
seem to have been discounted.

David




________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________