Re: ISSUE-4: Versioning, namespace URIs and MIME types ISSUE-60

Robert J Burns writes:

> identifying any conflicts ... name collisions where the same name
> could represent multiple different semantics and no context provided
> any means to discern those semantics.

Hi Rob.  I agree with your analysis that XHTML 1 <img> and XHTML 2 <img>
don't conflict when given the context of what attributes or content they
have.

What I'm still uneasy about is whether it's acceptable to force a
user-agent to have to use context clues like this to determine which
language it is dealing with.

If it is then we should make sure that, so far as XHTML 5 is involved,
this requirement is clearly stated somewhere, with the algorithm to be
used.  This could be something like 'Initially try parsing the content
as XHTML 2, and validate it; if it isn't valid then try again treating
it as XHTML 5.'

Or perhaps some cleverer algorithm could be employed, something like
'Initially keep an open mind, parsing as a quantum superposition of
"XHTML 2 or XHTML 5"; do not start to render or otherwise process the
document while the superposition still has multiple states.  If an <img>
element is encountered with an alt attribute then collapse the
superposition to XHTML 5; if an <alt> element is encountered with child
content then collapse the superposition to XHTML 2.  If the end of the
content is reached and the superposition still has multiple states then
pick any one.'

But it would hurt interoperability if different user-agents were to use
different algorithms, so this would need defining.

Smylers

Received on Wednesday, 18 February 2009 19:42:59 UTC