Re: handling XLink deprecation

Chris Lilley:

Thanks for the pointer.  Looks like my proposal was similar, but not the 

> CM> But I propose we:
> CM> * remove xlink:actuate, xlink:type, xlink:role, xlink:arcrole and
> CM> xlink:title altogether
> Yes
> CM> * change xlink:href to href for: <a>, <altGlyph>, the animation
> CM> elements, <color-profile>, <filter>, <font-face-uri>, <glyphRef>,
> CM> gradient elements, <mpath>, <pattern>, <textPath>, <tref> and <use>
> Yes (I already made that change for color-profile yesterday. Note though that @color-profile uses src; maybe color-profile should as well).
> CM> * change xlink:href to src for: <cursor>, <feImage>, <image> and <script>
> yes

For this, the above link says that we decided not to use src="" but to 
use href="" for these, and lists a few problems:

   * how does src work with respect to href or xlink:href
       * which attribute value takes precedence?
       * what is the name of the DOM attribute?
       * if the DOM attribute is set for one, is is changed for all?
   * will this confuse authors who expect <svg:image> and <html:img> to
     work the same?

Precedence is an issue even with xlink:href="" versus href="", so src="" 
doesn't introduce any new problem there.  (And we're not going to be 
allowing all three on one element.)

For the IDL attribute, I suggested in my earlier mail to have one called 
"src" that is a DOMString.  For whether setting it should affect both 
content attributes, I don't think so -- just make .src reflect affect 
only the src="" attribute.

I don't think this would increase confusion for authors any more than 
they already might be due to differences between HTML's <img> and SVG's 

Also on the wiki it is noted that HTML is not consistent with its use of 
href="" vs src="" for outbound vs inbound links.  But I think it's not 
inbound vs outbound that is the delineation -- it's whether the 
referenced content is included and rendered.  At least, that is one way 
of thinking about it that makes it seem more consistent.  (That ignores 
<object> having data="" of course!)

Received on Wednesday, 22 August 2012 00:42:01 UTC