Re: [xml-dev] Pete's blend for XMLlite

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

On 08/12/2010 15:09, Pete Cordell wrote:
> I think all of the points below are affected by this.

the ones relating to allowing -- in comments are not [in danger of 
corrupting data] as -- (including nested comments) are currently not 
well formed. Making currently non well formed stuff well formed is the 
safest incompatible change that could be made, followed by making well 
formed stuff non well formed (this is probably what killed xml 1.1) and 
then the most dangerous: making well formed stuff parse differently.

Personally if the final format specified doesn't have the property that 
every well formed xmmllite document is well formed xml 1.0, then I would 
arrange the syntax to ensure that _no_ well formed xmllite document is 
well formed xml 1.0, for example by mandating a declaration such as 
<?xmllite?> at the top.

The impact on "parsers" or "applications" are really not the main 
concern: it's the impact on "documents" that matters. The fact that (for 
example) the change in xml 1.1 to disallow C1 control characters was 
almost certainly a trivial change in any parser implementation wasn't 
really the point, the problem was that it made non well formed an 
unknown number of existing documents, so for most people it is far safer 
to stick with xml 1.0 than to move to xml 1.1 and potentially have the 
system fail at some point in the future when an old document is accessed.

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. 
________________________________________________________________________