Re: ISSUE-201: Aligning the two change proposals

Ted,

I see some gaps:

1. We need a section that defines how hit testing is processed in the HTML5
spec. Frank had such a process in the canvas proposal submittal that stated
the order to which hit testing was done based on the last drawn which dealt
with situations where hit regions overlapped. Without this developers will
not know how hit testing is handled by the browser. This should also make
a statement that when a hit is occurs (using the pointInRegion) that the
event is dispatched to the associated DOM element.
2. We need a method to remove all hit regions such as removeHitRegions() or
clearHitRegions()
3. Lightweight objects that provide only a label and a role are inadequate
to represent an object. For example, if you have role="checkbox" it can
never be checked. Also, I see no definition of how one creates a
lightweight object. If these are still in I don't see them as having been
adequately defined.

Rich


Rich Schwerdtfeger



From:	"Edward O'Connor" <eoconnor@apple.com>
To:	public-html@w3.org,
Date:	06/29/2012 05:06 PM
Subject:	Re: ISSUE-201: Aligning the two change proposals



Hi all,

I wrote:

> Frank and I got together today and went through the remaining
> ISSUE-201 items.

I've made a pass at updating my Change Proposal to reflect what came out
of that meeting. The updated proposal is available here:

         http://www.w3.org/html/wg/wiki/User:Eoconnor/ISSUE-201

Please review this updated proposal and let me know if I missed
anything. Here's the diff from the previous version of the proposal:

http://www.w3.org/html/wg/wiki/index.php?title=User%3AEoconnor%2FISSUE-201&diff=13191&oldid=12969


>> [Frank's CP only] A method ( clearElementPath() ) to remove hit
>> regions. I think we should have this in the joint proposal, as relying
>> on ClearRect or other drawing mechanisms to clear an association seems
>> overly involved and a burden on the author.
>
> We both agreed that having a clearElementPath() method makes sense.

To keep the method names aligned, I've added removeHitRegion() to my
proposal.

>> r7025 - Add ellipse support to canvas. not needed, not accessibility
> Agreed; can wait.
>
>> r7026 - Add SVG paths to Path objects in canvas.
> Agreed, can wait.

I removed these revisions from the list of revisions to restore.

>> r7028 - add dashed lines and change how Path objects work to instead
>> use external line and font styles and transformation objects
>
> I need to look at this in more detail to see if the refactoring is
> necessary to make Path sufficiently useful.

I think leaving this in is probably necessary to keep the Path
definition sane.

>> r7033 - More font metrics.
> Definitely helps wit hit testing runs of text, so I'd prefer to keep
> this in. That said, I can live with putting it off until HTML.next.
>
>> r7034 - Path copy constructor
> I don't see the harm in keeping it, but like the font metrics, I can
> live with putting it off as well.

I removed these revisions from the list of revisions to restore.

Rich wrote, in
<OF2EA82CD1.735CE9CF-ON86257A2A.006A8BD4-86257A2A.006ADC8C@us.ibm.com>:

> Does this mean that the ligtweight JSON objects will be an html.next
> discussion item?

Sorry, I failed to mention unbacked region descriptions in my summary
email of the conversation I had with Frank. I argued for keeping them in
this version of the feature. I think Frank could live with that, but I
don't want to put words in his mouth. Frank, what do you think? I
haven't changed this aspect of my proposal.


Ted