Re: [Jack-Devel] The Situation(s) With JACK
On 12/10/2011 12:58 AM, Paul Davis wrote:
> On Fri, Dec 9, 2011 at 6:47 PM, Florian Paul Schmidt
> <[hidden]> wrote:
>> Personally I found the jack transport specification always lacking in this
>> respect.
> lacking? you say its lacking but then ask for a very simple protocol:
Yes, because musical time is outside the range of what jack should IMHO
provide. What musical time is, and how it is shared between different
applications can be very application-specific.
>> IMHO a better approach would be to just privide a very simple
> which we have!
True :D
>> protocol, which is just framebased and provides just the states STOPPED,
>> RUNNING and a shared frame number. Making assumptions about musical time,
>> etc., is too restrictive.
> STOPPED and RUNNING can't accomodate slow-sync clients.
That's true. Here's a different approach maybe: Make clients share a map
of locations that can be relocated to (Slow sync clients could prebuffer
their state, so relocations/loops can be immediate). Make the update of
this shared map non-RT, but require clients to be able to relocate to
any of the shared locations immediately.
My point is: How to implement looping/musical time/etc., is for me still
subject to discussion and therefore should not be cast in stone in the
jack spec. Rather I'd like to see an independent spec to emerge for this
and once it is ready and well tested, then one can think about
reintegrating it into jack itself..
I have the feeling that the looping stuff and the musical time stuff are
instances of a more general problem: How can jack apps share state
besides the data passed around via ports? If there was a general
mechanism for jack apps to share (broadcast/pull) additional state, then
apps that want to provide notions of musical time or looping, etc, could
use this mechanism to implement this. Providing a playing ground for
experiments with this kind of things..
Maybe i just need more coffee though ;D
>> To provide musical time interoperation between different audio apps an
>> independent protocol could emerge and once it has matured one could again
>> think about integrating it into jack transport..
> we've had the simple protocol for years now. we've also had MMC for
> even longer, which is totally independent of JACK but has been
> integrated into via jackmmcctl or whatever its called.
>
> i recommend reading the wiki page cited on this. it has an excellent
> discussion of the entire issue.
OK, will do so. Thanks for the pointer..
Flo
1323519887.3573_0.ltw:2,a <4EE34F8A.4010407 at gmx dot net>