Re: MathML in HTML: syntax for hypertext links

On Apr 3, 2009, at 12:16, David Carlisle wrote:

>> Actually, the parsing side of xlink:href in HTML5 is a solved  
>> problem.
>
> Henri, sorry if this is a silly question, but do you mean it's  
> solved in
> Gecko or solved in the html5 specification?

Solved in the HTML5 spec and on the parser layer in HTML5 parser- 
enabled Gecko builds. However, the above-DOM linking behavior is  
broken, so it doesn't help much yet that the parser part is working.

> And aside from any hopefully temporary problems with xlink in mathml  
> in
> Gecko. Have you any thoughts on whether we should allow href  
> everywhere
> in mathml.

Introducing a duplicate-namespace mechanism for a thing that already  
had a mechanism turned out to be a very bad idea in the case of id and  
xml:id. I hear that the issues with xml:id are now recognized in the  
SVG WG. I wonder if the Math WG could avoid the problem entirely and  
not introduce xml:id into MathML at all.

I dislike XLink and Namespaces, but my initial reaction is that  
introducing a duplicate mechanism is trouble compared to defining  
xlink:href semantics on MathML (and SVG) nodes independently of XLink  
and zapping generic XLink from browsers. Linkness is very different  
from IDness, though, so I'm not sure if having a duplicate mechanism  
for xlink:href in MathML (and SVG) is as bad as having a duplicate  
mechanism for id. A duplicate mechanism for xlink:href might be  
harmless, but I'm not ready to say so yet, because I haven't thought  
about it carefully and I'm predisposed to assume it's not harmless.

> Was for example the xhtml2 style of href on all html elements  
> considered
> and rejected for html5, or has it just not come up?

Considered and rejected. (It comes up very often.)

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Friday, 3 April 2009 09:45:15 UTC