Re: WebGL | The 3D Canvas Context for HTML

On Fri, 11 Dec 2009 23:25:02 +0100, Arun Ranganathan <arun@mozilla.com>  
wrote:
> Anne van Kesteren wrote:
>> https://cvs.khronos.org/svn/repos/registry/trunk/public/webgl/doc/spec/WebGL-spec.html  
>> The draft appears to use outdated Web IDL syntax. It also uses  
>> features, e.g. NameDeleter that are controversial with TC39 so you may  
>> want to reconsider using those.
>
> Can you give any specific syntax that's outdated?  Also, are you  
> referring to the published WebIDL specification [1] or the editor's  
> draft [2]? Feedback about NameDeleter is useful.

I was referring to the editor's draft. Didn't realize the new syntax for  
e.g. getters and setters hadn't made it onto TR/ yet. (It has made it into  
drafts I edit :-))


>> It also seems that a bunch of the new objects introduced should really  
>> be part of ECMAScript instead. E.g. all the new typed array interfaces.
>
> The goal would certainly be to feed some of this back to ECMAScript.

So maybe IDL syntax should be used that makes sure that implementations  
hide the interfaces from the global object so applications do not start to  
rely on them. That way we can change things (i.e. unify) more easily going  
forward.


>> Some of the typedefs are scoped to WebGLRenderingContext but are in  
>> fact used all over the place.
>
> This should be fixed; can you share a specific one?

GLint for instance.


>> Also, why is it WebGLRenderingContext and not  
>> CanvasRenderingContextWebGL which would be more consistent with the  
>> existing CanvasRenderingContext2D?
>
> 1. Length of typing :)

You will almost never have to type this interface name as developer.


> 2. You'll notice that the nomenclature is such that *all* interfaces are  
> prefaced with WebGL*.  So consistency with the other interfaces is  
> desirable.  This naming convention was more convenient to JS developers,  
> rather than OGL ES 2.0 "gl prefixing."

It would still make more sense to me if the rendering context interface  
was named in a way consistent with the existing rendering context  
interface. It is in that way somewhat different as well from the other  
interfaces introduced.


> [1] http://www.w3.org/TR/WebIDL/
> [2] http://dev.w3.org/2006/webapi/WebIDL/


-- 
Anne van Kesteren
http://annevankesteren.nl/

Received on Wednesday, 16 December 2009 10:14:39 UTC