4 Oct
2002
4 Oct
'02
8:27 a.m.
On Fri, 4 Oct 2002, Steve Harris wrote:
This might be a good addition. In any case using pids is not optimal.
This is basicly the same problem as with port names. Either we maintain
two sets of names (one for programs, one for the user-interfaces) or
make a compromise and have something that is usable for both. Requiring
users to map the application instances to pids does not fill this
requirement.
Having a basename with a standard way of labelling/iterating instances
is the way to go IMHO. Ecasound does the following now:
--cut--
for(n = 0; n < AUDIO_IO_JACK_MANAGER::instance_limit; n++) {
client_repp = jack_client_new (client_name.c_str());
if (client_repp != 0) break;
client_name = jackname_rep + "_" + kvu_numtostr(n + 2);
}
--cut--
Ie. it tries "ecasound", "ecasound_2", and so on. The exact syntax
doesn't
really matter, just as long as all client apps use the same one.
--
http://www.eca.cx
Audio software for Linux!
I'm not
sure that you'd want to do this with the meterbridge but it's
something that needs to be handled if we want to be able to save and restart
a "jack session".
I dont like using pids because it looks messy, maybe
jack could provice an
eqivalent to tmpname (or whatever its called), so you could call
jack_unique_name("bridge") and get back "bridge-1" the first time
and
"bridge-2" the second for example.