Re: WaveShaperNode.curve

On Mon, May 13, 2013 at 12:45 PM, Ehsan Akhgari <ehsan.akhgari@gmail.com>wrote:

> The downside of that approach is that the UA will accept code which
> doesn't do what they intended, such as:
>
> shaper.curve = new Float32Array(100);
> for (var i = 0; i < 100; ++i) {
>   // oops, at this point the array has been internally copied,
>   // and while the code below doesn't throw an exception,
>   // it effectively sets the curve property to an all-0 array,
>   // which is not what the author has intended.
>   shaper.curve[i] = whatever;
> }
>
> I think this would be a terrible API.
>

Agreed, that's a good point.   Then lets not copy the array.


>
> For reusing the ArrayBuffer, the author can just read the curve property
> again and get a copy back which they can use for other purposes...
>
> --
> Ehsan
> <http://ehsanakhgari.org/>
>
>
> On Sun, May 12, 2013 at 10:16 PM, Chris Rogers <crogers@google.com> wrote:
>
>>
>>
>>
>> On Sun, May 12, 2013 at 2:20 PM, Ehsan Akhgari <ehsan.akhgari@gmail.com>wrote:
>>
>>> The current WebKit implementation of this node is racy, since the
>>> processing code only protects against simultaneous setting of the curve
>>> property, not against modifying the contents of the ArrayBuffer.
>>>
>>> In the Gecko implementation, I'm just copying the contents of the array
>>> upon setting curve for now, but I think a better fix would be to neuter the
>>> contents of the array, and provide a copy of the original contents of the
>>> array if contents reads the curve property again.
>>>
>>> Does this make sense?
>>>
>>
>> I much prefer an internal copy, and that can even be optimized as a fast
>> pointer swap.  I don't like the idea of harming the ArrayBuffer so that it
>> can't be used again.
>>
>>
>>>
>>> Thanks!
>>> --
>>> Ehsan
>>> <http://ehsanakhgari.org/>
>>>
>>
>>
>