Re: [WebIDL] prototype chains, multiple inheritance, mixin interfaces

Andrew Oakley:
> > Is this a reasonable way of dealing with the problem?  Are there any
> > issues with doing this?  Is it any better than the proposal from the
> > TC39 meeting?

Brendan Eich:
> C3 is reasonable in my opinion, but it's strictly more complicated
> than single inheritance with mixins implemented by "copying" (or
> as if they are macros). If JS developers do not need reified mixin
> interfaces with .prototype properties, then simpler wins. That was our
> judgment last week at the TC39 meeting.

Right.  Ages ago the spec defined even more complicated handling of
multiple inheritance (but without C3 linearisation) than it did recently
with the introduction of the mixin prototype object.  Even with the
proposal to use C3 linearisation rather than whatever perhaps
ill-defined one was in the spec back then, I got mostly negative
feedback because (1) it required custom [[Get]] etc., (2) it was
complicated, and (3) it would likely result in poorer performance.

On the positive side, it got you the ability to monkeypatch abstract/
mixin interfaces.

I think these are the tradeoffs, and what I heard was that the simpler
single inheritance mechanism as proposed, especially in combination with
making EventTarget a real ancestor of Node, XMLHttpRequest, etc., was
much more acceptable to implementors – especially since it is closer to
what they all (kinda differently) do at the moment.

-- 
Cameron McCormack ≝ http://mcc.id.au/

Received on Tuesday, 23 November 2010 22:57:28 UTC