Re: [Sc-devel] [RFC] sample accurate scheduling

[blackrain <blackrain.sc@xxxxxxxxx>]

On a second thought, yes, that approach should work in theory but
again, sample rate fluctuations in the hardware will cause the effects
we see now - even if we schedule packets based on SampleClock units of
time and state deltas based on that, we can not be sure the delta will
be accurate by the time it reaches the server which will reinterpret
the packet's time stamp under its current hostime stamp scheme.

IMO we need a sample based schedule model to match SampleClock.

x

On Dec 1, 2007 6:00 PM, blackrain <blackrain.sc@xxxxxxxxx> wrote:
> On Dec 1, 2007 1:09 PM, Jan Trutzschler <sc3@xxxxxxxxxxxxxxxxx> wrote:
> > I think Scott is right. I think the server should only have the
> > option to run its scheduler on the hardware's sample clock.
> > j
>
> this could be yet another approach (if I understand right).
> have scsynth 'main' clock be the hardware's sample clock and translate
> osc packets timestamps to that - it sounds like quite some work and a
> bit risky.
> I believe keeping the scheduling model we have and implementing
> SampleClock in the lang side is a more straight forward approach.
>
>
> x
>
> > On Dec 1, 2007, at 6:32 PM, Scott Wilson wrote:
> >
> > > Why not simply use timestamps in seconds and then maintain OSC
> > > compatibility rather than implement a new message and queue? If the
> > > clock is the same everything should align correctly, yes?
> >
> >
> > _______________________________________________
> > Sc-devel mailing list
> > Sc-devel@xxxxxxxxxxxxxxx
> > http://www.create.ucsb.edu/mailman/listinfo/sc-devel
> >
>