Re: Missing information in the Web Audio spec

On Fri, May 11, 2012 at 4:38 AM, Raymond Toy <rtoy@google.com> wrote:

> On Wed, May 9, 2012 at 6:19 PM, Robert O'Callahan <robert@ocallahan.org>wrote:
>
>> In AudioParam, a lot of the processing model is unclear. I assume that
>> nominally an AudioParam is a function from time to floats. So, for
>> example,what does setValueAtTime actually do? Does it set the value for all
>> times >= 'time'? Does setting the 'value' attribute make the function
>> constant over all times? And what does it mean to be "relative to the
>> AudioContext currentTime"? Does that mean passing 0 changes the value at
>> the current time, or that 'time' and 'context.currentTime' are simply on
>> the same timeline? If the former, clarify by saying that 0 corresponds to
>> the context's currentTime.
>>
>
> Heh.  I thought setValueAtTime was the easiest method to understand.  It
> sets the value of the AudioParam to the specified value at time T, where T
> is the offset from context.currentTime.  Unless you do something else to
> it, it stays at this value.
>

But it's got to change the value over some time interval, not just for some
instant. So what is that time interval? Until the next scheduled change?
What exactly is the "next scheduled change", for example if setValueAtTime
is called for a time that's in the middle of an interval defined by a
previous setValueCurveAtTime?

Yes, this could probably be added, although I think linearRampToValueAtTime
> is pretty clear.
>

It's actually not. What is the time interval over which the linear ramp
happens? From the "last scheduled change"? Like above, "last scheduled
change" needs to be defined. For example, does the "last scheduled change"
consider only changes that are already scheduled when
linearRampToValueAtTime is called? If not, how and when is "last scheduled
change" evaluated?

exponentialRampToValueAtTime is less so because there was some issue raised
> that you can't start or end with a negative value (because exp(x) is always
> positive for real x).
>

What is the exponent? If it's "e" (Euler's number), that needs to be stated
in the spec.

setValueCurveAtTime seems straighforward, unless you need to describe how
> the interpolation is done if the number of values doesn't exactly match the
> duration.
>

"number of values" and "duration" aren't even the same type, so we can't
talk about whether they "match". Certainly the exact method of
interpolating values must be defined. To avoid ambiguity, instead of
referring to "first order filter" or whatever, the spec should just give a
mathematical definition and describe the filter in non-normative text.
(Although honestly I don't know why we're referring to first-order filters
here when the author's given the parameter values explicitly.)

Rob
-- 
“You have heard that it was said, ‘Love your neighbor and hate your enemy.’
But I tell you, love your enemies and pray for those who persecute you,
that you may be children of your Father in heaven. ... If you love those
who love you, what reward will you get? Are not even the tax collectors
doing that? And if you greet only your own people, what are you doing more
than others?" [Matthew 5:43-47]