On Tue, 2006-11-28 at 15:58 +0200, Mikko Rantalainen wrote:
> Is there a way to get correct baseline alignment when MathML is 
> placed inside an OBJECT? Yes, it's possible to get somewhat 
> acceptable result but HTML+OBJECT isn't a substitute for XHTML+MathML.

I tried Googling for the problem you refer to, but couldn't find
anything. Do you have a link or example of a baseline problem in OBJECT?

> OK. Usually one of the serialized formats is the "native" format but 
> not in all cases.

I'm not sure this is usual anymore, with the proliferation of content
management and blogging software.

> If you follow my logic above that there's a *source* format for the 
> document and user is allowed to fetch that version then it would be 
> the "source variant". As it should have the best quality it should 
> be transmitted if Accept's "*/*" media type has quality of 1 or 
> Accept header does not exist.

While I think we can certainly improve even installed UAs to send more
helpful Accept headers, but it may be overambitious to assume we can
change /all/ installed UAs to stop mindlessly sending */*. So without a
HTTP 1.2 to differentiate old from new style requests that would be
problematic. (I suppose alternatively we could actually use a completely
new header like Accept-For-Rendering or something.)

> I would expect a normal UA to send list of media types it "supports" 
> in its normal Accept header but there could be an UI to fetch the 
> "source" version of the document where the UA would repeat the 
> request without Accept header (or with Accept: */*).

Sorry, Mikko, I'm still not clear what you mean by "supports". And would
you mind elaborating on how such a user interface might work and what it
might look like in practice (from the end-user's perspective I mean)?

--
Benjamin Hawkes-Lewis