[whatwg] Canvas and imageData

[Ian Hickson]

n Tue, 15 May 2007, Maciej Stachowiak wrote:
> > > 
> > > Also, if it's meant to be required for implementations to allow 
> > > that, the spec should say so - it's not normally assumed that any JS 
> > > object with the right properties can be used anywhere that an 
> > > interface can be used.
> > 
> > Isn't it?
> 
> I wouldn't expect something like the following to work, and indeed it 
> does not in any browser I tried. I would not a JS object with all the 
> attributes and methods of the core DOM Text interface could be inserted 
> into the DOM and render either.
> 
> <div id="container">
> <div id="test">test div</div>
> </div>
> 
> <script>
> var myEvent = { type: "custom", target: document.getElementById("test"),
> currentTarget: document
>               , eventPhase: 0, bubbles: true, cancelable: false, timeStamp:
> Date()
>               , stopPropagation: function() { }, preventDefault: function() {
> }
>               , initEvent: function(eventTypeArg, canBubbleArg, cancelableArg)
>                 { this.type = eventTypeArg; this.bubbles = canBubbleArg;
> this.cancelable = cancelableArg; } };
> 
> document.getElementById("container").addEventListener("custom",
> function(event) { alert("event"); }, true);
> document.getElementById("test").dispatchEvent(myEvent);
> 
> </script>

It seems to me based on this:

   http://www.w3.org/TR/2003/NOTE-DOM-Level-3-Events-20031107/events.html#Events-CustomEvent

...that at least some people expected that kind of thing to work.

However, that's removed from the current DOM3 Events draft, so, ok.


> It is typical for browsers to only accept objects that have the right 
> custom internal implementation, or in JS terms, the right value of the 
> [[Class]] internal property, as paramaters to DOM methods requiring a 
> DOM interface, except in special cases.

Hm, valid point. I've added that as a requirement, and made ImageData 
explicitly override it.


> > On Mon, 14 May 2007, Anne van Kesteren wrote:
> > > 
> > > Just to keep this list in sync with #whatwg. It was suggested to 
> > > give both putImageData and getImageData a "high resolution" boolean 
> > > parameter which would indicate what type of ImageData object you 
> > > would get back. This would be fine by me. I'm not sure if that's 
> > > needed right away though.
> > 
> > That's another option, indeed. We could address this when someone 
> > implements a high-res <canvas>.
> 
> Done. The WebKit <canvas> implementation scales the backing store by the 
> Mac OS X system-wide UI scale factor. The code to create the scaled 
> backing store starts on line 186 of WebCore/html/HTMLCanvasElement.cpp. 
> See http://nightly.webkit.org for a downloadable copy. You can use 
> Quartz Debug.app to set the UI scale factor to something other than the 
> default value of 1.0. (Apple does not currently ship any systems with 
> the scale factor pre-set to anything but 1.0.)

Do you have any experience with getImageData() and putImageData() that 
would help us here?

I couldn't find any running non-demo code to test. If it turns out that 
browsers doing this ship soon, this may become a non-issue, since authors 
will find things break as soon as they try doing it wrong.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'