On 05/04/2011 02:48 PM, Fons Adriaensen wrote:
...
> The only real bug is that this doesn't
work correctly with
> the optimisation for a single output connected to an input.
exactly.
if you connect any other thing to that same input port the "zero-copy"
issue goes away :)
side-note to who might concern: it has been previously reported as a
qtractor bug though, when doing this kind of direct loopback in a
brainless bouncing scenario. fact is, it's one half of the problem (jack
zero-copy optimization)--the other half gets to which track bounces to
which first (ie. depends on process/track order) and that's definitively
a qtractor's under-the-hood issue but not one i call urgent atm.
besides, jack's "zero-copy" issue remains still, attotw
alas, i am not that convinced it is a jack bug. if one, say a jack
client, ever wants to allow a direct loopback, why doesn't he/she do the
appropriate code him/her-self? it's a lousy memcpy() away isn't it?
byee
--
rncbc aka Rui Nuno Capela
rncbc(a)rncbc.org