Re: JS 45 problem

Thanks a lot.  I remember talk about this. Luckily we don't (yet!) have 
a lot of code using JS_InitClass

On 05/24/2016 09:47 AM, Boris Zbarsky wrote:
> On 5/24/16 10:32 AM, Kent Williams wrote:
>>         obj = JS_NewObject(cx, &TESTO);
>
> Kent,
>
> This line is the problem.  JS_NewObject used to guess at the prototype 
> to use based on the name of the given class and what that name 
> resolved to in global scope.
>
> This was all sorts of broken and fragile, and was removed (see 
> <https://bugzilla.mozilla.org/show_bug.cgi?id=1125567>).  If you use 
> JS_NewObject, you will get an object whose prototype is Object.prototype.
>
> What you want here is JS_NewObjectWithGivenProto and pass in the thing 
> you want as your prototype.
>
> I think you have a few options for how to do that:
>
> 1)  Save the value returned from JS_InitClass and use that. Should be 
> the right thing as long as no one is subclassing you.
>
> 2)  In the constructor, get the "prototype" property of args.callee() 
> and use that.  Again, should be the right thing as long as no one is 
> subclassing you.
>
> 3)  In the constructor, get the "prototype" property of 
> args.newTarget().  Should do the right thing (as in, use the proto of 
> the constructor that was actually invoked) even if you're subclassed.
>
> -Boris
> _______________________________________________
> dev-tech-js-engine mailing list
> [email protected]
> https://lists.mozilla.org/listinfo/dev-tech-js-engine