Constructors without new
I'm guessing this is one of the things you needed to file a bug for? Considering the last draft was published before the consensus (throw if not "new"), I think this just need to be updated :)
On Aug 7, 2013, at 4:14 AM, Anne van Kesteren wrote:
Per last meeting I understood that constructors without new should probably not work to make subclassing easier (my understanding of that is still somewhat limited, I'll get there). I filed www.w3.org/Bugs/Public/show_bug.cgi?id=22808 on IDL to see if this was possible for the platform side of things.
I now see that ArrayBuffer() is defined as doing the equivalent thing to new ArrayBuffer() in the ES6 draft, despite e.g. Chrome currently throwing for the former. (This is equivalent to today's story for e.g. XMLHttpRequest.)
Actually, the algorithm in the current (Rev16) draft does not treat these as equivalent. ArrayBuffer( ) will thrown according to the algorithm.
However, there is an issue in Rev16 in that the prose description in 15.13.5.still said that ArrayBuffer ( ) and new ArrayBuffer( ) are equivalent. That's simply an editorial bug that has already been corrected in my working draft for Rev17.
A good example, of why I try to minimize redundant prose descriptions and why reader should always use the algorithms as the primary definition.
Per last meeting I understood that constructors without new should probably not work to make subclassing easier (my understanding of that is still somewhat limited, I'll get there). I filed www.w3.org/Bugs/Public/show_bug.cgi?id=22808 on IDL to see if this was possible for the platform side of things.
I now see that ArrayBuffer() is defined as doing the equivalent thing to new ArrayBuffer() in the ES6 draft, despite e.g. Chrome currently throwing for the former. (This is equivalent to today's story for e.g. XMLHttpRequest.)
What exactly is the plan here?