Re: [geometry] DOMRectList with legacy

On Jul 18, 2014, at 8:20 AM, Rik Cabanier <cabanier@gmail.com> wrote:

> 
> 
> 
> On Thu, Jul 17, 2014 at 11:10 PM, Dirk Schulze <dschulze@adobe.com> wrote:
> 
> On Jul 18, 2014, at 7:05 AM, Boris Zbarsky <bzbarsky@MIT.EDU> wrote:
> 
> > On 7/17/14, 5:37 PM, Dirk Schulze wrote:
> >> With the current limitations of WebIDL, does it mean you prefer using [ArrayClass]? (Instead of subclassing Array?).
> >
> > For green-field things, I would prefer just using an Array for now, then doing subclassing once it's implemented.
> >
> > For legacy stuff, subclassing Array is not an option thus far, because none of the JS engines support subclassing yet.  This has nothing to do with WebIDL limitations; it's a V8/SpiderMonkey/Chakra/JSC limitation.
> >
> >> Does that work together with [NoInterfaceObject]
> >
> > You mean [ArrayClass]?  Yes, it does.
> >
> >> and would you even consider it?
> >
> > Consider [NoInterfaceObject], you mean?  I could probably live with that, yes.
> 
> Thanks Boris! Anne, Domenic, Ms2ger, Rik does that sound like a compromise:
> 
> [NoInterfaceObject,
>  ArrayClass]
> interface DOMRectList {
>     readonly attribute unsigned long length;
>     getter DOMRect? item(unsigned long index);
> };
> 
> Weren't we going to rename it to LegacyDOMRectList so people don't try to reuse it in the future?
> I would still like to know if other browser are onboard with implementing [ArrayClass] and extending this object with that feature.

There is a huge normative note that it should not be used (getClientRects() will obviously be the exception). Since we don’t spread the name in the global namespace because of the [NoInterfaceObject], I see all concerns from Anne and Domenic addressed.

Greetings,
Dirk

Received on Friday, 18 July 2014 08:08:55 UTC