[Jack-Devel] Odp: Re: Odp: Re: jack1: jack_transport does not work as master

I send a patch once on this list and this is a "patched" version.
I keep this, because it was very helpfull while testing and improve other apps ( like hydrogen , fsthost ), especially tempo change issues. Not perfect but better than "official", especially in low tempos ( like 30 BPM ).
.. so yes you are inconvincible ;-)

P.

Dnia Czwartek, 2 Lipca 2015 20:17 Paul Davis <[hidden]> napisa³(a) 
> Was this the stuff you tried (unsuccessfully) to convince me need to
> go into the tools repo?
> 
> On Thu, Jul 2, 2015 at 1:12 PM, Pawel <[hidden]> wrote:
> > Hi,
> >
> > Tobias, you can check my slightly improved version of jack_transport:
> > https://sourceforge.net/p/nxjsm/code/HEAD/tree/trunk/tools/transport.c
> >
> > I'm also the author of new Hydrogen jack-master code. It's works much better , but probably not perfect ;-)
> >
> > P.
> >
> > Dnia Czwartek, 2 Lipca 2015 18:57 Paul Davis <[hidden]> napisa³(a)
> >> Frankly, computing BBT time has never been part of "how to use the
> >> JACK API". It should probably never have been included in the source
> >> code in the first place.
> >>
> >> Managing BBT time is *extremely* complex - we have worked on it for
> >> years in Ardour and it still has issues. I'd prefer to just rip it
> >> out.
> >>
> >> On Thu, Jul 2, 2015 at 12:47 PM, Tobias Hoffmann
> >> <[hidden]> wrote:
> >> > On 02/07/15 17:30, Tobias Hoffmann wrote:
> >> >>
> >> >> On 02/07/15 15:38, Paul Davis wrote:
> >> >>>
> >> >>> jack_transport is not intended to act as a the time master.
> >> >>
> >> >> Well, comparing example-clients/transport_client.c and tools/transport.c,
> >> >> I'd say the one big difference between the two is that tools/transport.c
> >> >> (aka jack_transport) can act as a *simple* timebase master...
> >> >
> >> >
> >> > Ok, I see now that the BBT update algorithm (in the "else" branch in
> >> > timebase()) in transport.c is quite broken, because it accumulates rounding
> >> > errors. However always using the first, recalculating branch (basically: if
> >> > (true)) seems to be stable -- or at least: better than Hydrogen's timebase
> >> > master, which updates the BBT only every few cycles... (I've read that
> >> > somewhere a completely rewritten Jack Master implementation for Hydrogen is
> >> > available, which I have not tested yet).
> >> >
> >> > I find it quite sad that even some code distributed in the jack tarball is
> >> > an example of how NOT to use the Jack API ...
> >> >
> >> >   Tobias
> >> >
> >> >
> >> >
> >> >
> >> >
> >> 
> >> Jack-Devel mailing list
> >> [hidden]
> >> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
> >
> >
> >
> > 
> > Jack-Devel mailing list
> > [hidden]
> > http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

1435866688.22432_0.ltw:2, <5595963424e900.26629329 at wp dot pl>