Sorry I didn't send Stéphane's more general code - thanks for stepping in.
On Tue, Oct 9, 2018 at 2:12 AM Stéphane Letz <l...@grame.fr> wrote:
>
>
> > Le 9 oct. 2018 à 10:42, Oleg Nesterov <o...@redhat.com> a écrit :
> >
> > On 10/08, Julius Smith wrote:
> >>
> >> Very nice! Stéphane's solution is to add the needed state in a
> >> subclass of dsp defined in the architecture file. See the attached
> >> simple example ttfs = "test ffunction with state".
> >
> > Heh. Thanks, now I understand what you mean.
> >
> > This is very inconvenient and limited. And btw, this example doesn't look
> > right.
> >
> > import("stdfaust.lib");
> >
> > ff_smooth = ffunction(float ff_smooth(float,float),"","");
> > ff_init = ffunction(float ff_init(float),"","");
> >
> > smoothness0 = 0.5;
> > smoothness1 = 0.9;
> > smoother(smoothness) = ff_smooth(smoothness) : attach(ff_init(ma.SR));
> > smoother0 = smoother(smoothness0);
> > smoother1 = smoother(smoothness1);
> >
> > but this will not create 2 smoothers, class State has a single "float state"
> > initialized twice in a row, and then
> >
> > process = 1-1' <: smoother0 , smoother1;
> >
> > ff_smooth() will be called twice too; the smoothness/x args are correct but
> > again, both calls will play with the same "state".
> >
> >
>
> Well the example with 2 states was actually this code, with explicit 2 State1
> and State2 classes…etc..
>
>
>
>
>
> > I modified tffs.dsp:
> >
> > smoother = FPP(smoothness, x) {
> > // shared by all instances, this is only needed to create the
> > // unique "$smoother" id printed below.
> > DECL: int smoothers;
> > INIT: smoothers = 0;
> >
> > // every instance has its own state/smother
> > decl: int $smoother;
> > float $state;
> >
> > init: $state = 0;
> > $smoother = smoothers++;
> >
> > std::cout << "created smoother " << $smoother << std::endl;
> >
> > exec:
> > $state = (1.0f - $smoothness) * $x + $smoothness * $state;
> > std::cout << "smoother " << $smoother << ": " << $smoothness
> > << ", x: " << $x << ", y: " << $state << std::endl;
> > $state;
> > };
> >
> > smoothness0 = 0.5;
> > smoothness1 = 0.9;
> > smoother0 = smoother(smoothness0);
> > smoother1 = smoother(smoothness1);
> >
> > process = 1-1' <: smoother0 , smoother1;
> >
> > then changed Makefile to run fpp instead of faust, and the output is
> >
> > created smoother 0
> > created smoother 1
> > smoother 0: 0.5, x: 1, y: 0.5
> > smoother 1: 0.9, x: 1, y: 0.1
> > smoother 0: 0.5, x: 0, y: 0.25
> > smoother 1: 0.9, x: 0, y: 0.09
> > smoother 0: 0.5, x: 0, y: 0.125
> > smoother 1: 0.9, x: 0, y: 0.081
> > smoother 0: 0.5, x: 0, y: 0.0625
> > smoother 1: 0.9, x: 0, y: 0.0729
> > smoother 0: 0.5, x: 0, y: 0.03125
> > smoother 1: 0.9, x: 0, y: 0.06561
> > smoother 0: 0.5, x: 0, y: 0.015625
> > smoother 1: 0.9, x: 0, y: 0.059049
> > smoother 0: 0.5, x: 0, y: 0.0078125
> > smoother 1: 0.9, x: 0, y: 0.0531441
> > smoother 0: 0.5, x: 0, y: 0.00390625
> > smoother 1: 0.9, x: 0, y: 0.0478297
> >
> > which looks correct.
> >
> > process = 1-1' <: par(i, 10, smoother((i+1)/10));
> >
> > outputs
> >
> > created smoother 0
> > created smoother 1
> > created smoother 2
> > created smoother 3
> > created smoother 4
> > created smoother 5
> > created smoother 6
> > created smoother 7
> > created smoother 8
> > created smoother 9
> > smoother 0: 0.1, x: 1, y: 0.9
> > smoother 1: 0.2, x: 1, y: 0.8
> > smoother 2: 0.3, x: 1, y: 0.7
> > smoother 3: 0.4, x: 1, y: 0.6
> > smoother 4: 0.5, x: 1, y: 0.5
> > smoother 5: 0.6, x: 1, y: 0.4
> > smoother 6: 0.7, x: 1, y: 0.3
> > smoother 7: 0.8, x: 1, y: 0.2
> > smoother 8: 0.9, x: 1, y: 0.1
> > smoother 9: 1, x: 1, y: 0
> > ...
> >
> > No need to create a subclass, no need to write the additional C++ code, etc.
> >
> > Oleg.
> >
>
> Julius initial demand was more limited, and the proposed solution was enough
> I think.
>
> OK your solution seems more general, so we are interested to see your FPP
> code ((-;
>
> Stéphane
>
--
Julius O. Smith III <j...@ccrma.stanford.edu>
Professor of Music and, by courtesy, Electrical Engineering
CCRMA, Stanford University
http://ccrma.stanford.edu/~jos/
_______________________________________________
Faudiostream-devel mailing list
Faudiostream-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/faudiostream-devel
