Re: [Jack-Devel] Patches for jack-0.121-3

On Mon, Mar 19, 2012 at 10:43:34AM -0400, Paul Davis wrote:
> Am I right that the presence of jack_get_cycle_times() should lead us
> to deprecate jack_last_frame_time() or at best make its implementation
> use jack_get_cycle_times()? otherwise it seems that we're just
> cluttering the API unnecessarily .. comments?

It's somewhat easier to use than jack_get_cycle_times(), and
for many clients it could be all they need. 
Since the implementation is basically just one function call
implementing it by using jack_get_cycle_times() would actually
complicate rather than simplify it.
The only gripe I have with jack_last_frame_time() is that it's
such a misnomer...

One real improvement could be to ensure that copying the timer
information (involving the 'ugly hack') is done only once in
each cycle by caching the result at the client side. The copy
would then need to be marked as invalid at the start of each
cycle.


There is one more fix needed in engine.c : when the buffer
size is modified, engine->first_wakeup should be set. This
will re-init period_usecs (which will otherwise take a long
time to catch up), and also reset the filter coefficient for
the correct DLL bandwidth.


Ciao,

-- 
FA

A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

1332169661.32542_0.ltw:2,a <20120319150724.GB26430 at linuxaudio dot org>