On 14/01/2011, at 15:10, Boris Zbarsky wrote:
> On 1/14/11 5:42 AM, Jorge wrote:
>> Are you sure there's really a "maybe" for methods ?
>> Methods' functions have access to (at least) the sender's global context and 
>> contexts can't be shared, how to deal with that if not by prohibiting 
>> methods ?
> 
> We might need some ES spec changes/additions to produce methods that would be 
> safe to pass.  I agree that it's not that feasible for now.
> 
> (Though with enough hackery it could all be done: the method could sync proxy 
> the property access on things that are not on the worker thread to the main 
> thread, etc.  It's probably not worth it.)

Perhaps not worth it.

>> WRT to prototypes, to keep it simple, I think .__proto__ should be null, 
>> because we can't pass (and make unreachable) things like Object.prototype :-)
> 
> So proto is null on the result of the clone; not required to be null on the 
> thing being cloned.  Right?

Yes, afaics, that would be fine, wouldn't it ?

That, or the requirement for it to be one of the standard [[prototype]]s : 
Object.prototype, Array.prototype, etc, so that it could be readjusted easily 
to point to its homologue in the receiving context/thread. (?)
-- 
Jorge.

Reply via email to