Re: well-formedness error

[Danny Ayers <danny666@xxxxxxxxxxx>]

Tim Bray wrote:

> ...recovery from broken headers feels less pernicious to me than 
> recovery from a missing end-tag.


Same here, especially if that's the current default implementation in 
HTTP+XML systems. But I think this approach could only be justified if 
it could be done without conflicting with existing specs.

If I'm reading things correctly, the dependency on responding correctly 
to the header is in RFC 2023, which is under revision. Perhaps there may 
be an approach available by which the specification of a particular 
application can override the default rules, and insist on looking at the 
doc post-delivery. In other words, is there/could there be a loophole?

Mandating utf-8 might make a simple pragmatic answer to the whole 
problem. utf-16 would be desirable too, but that's reintroducing the 
potential for conflict. I dunno.

The <Atøm> idea is rather elegant ;-)

Cheers,
Danny.

>


--

Raw
http://dannyayers.com