On 2014/11/30 21:41:56, caitp wrote:
On 2014/11/30 21:32:54, Dmitry Lomov (chromium) wrote:
> On 2014/11/30 20:32:08, caitp wrote:
> > As mentioned in the CL description, I believe the draft's behaviour
results
in
> a
> > quite observable difference in behaviour from the current
implementations
in
> > browsers.
> >
> > I haven't done any research to verify if that difference could break
> > applications or not, but it might be good to answer those questions
here.
> >
> > Also as noted, this does not modify the runtime concat routine which
appears
> to
> > be carefully optimized for previously spec'd behaviour.
> >
> > There are still some issues with the subclassed-array behaviour that
is
> > mandated. Maybe a good place to figure that out
>
> Frankly, not sure we want to lead the way for other implementation in
this
> particular case :)
> Risks seem great for not much benefit. Maybe we should go back to TC39
with
> this.
I was talking about it in #jslang, it looks like this behaviour has
actually
not
changed from ES5 (ES262 5.1 15.4.4.4 step 5.b.iii.3) --- however yes,
JSC, SM
and V8 are all violating this, it's not clear whether it's fixable or not.
So, I am just reading the spec wrong, jorendorff corrected me here. So
there's
no wildly web-breaking behaviour.
https://codereview.chromium.org/771483002/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
