Le 19 mai 09 à 11:29, Sampo Savolainen a écrit : > Quoting "Fons Adriaensen" <[email protected]>: > >> I would have no objection if you added e.g. >> >> jack_client_open_via_dbus() > > How is the application supposed to know whether the user wants to use > dbus or not? > >> leaving the original call as it is. > > I agree with Stephanes' 5): > > An implementation where dbus and "classic" coexist in a single build > is the only way to go. The user environment would somehow tell jack > whether the user wants to use dbus or not. (In fact, I'd argue this > path should've been chosen from the start.) > > Say for example "the file ~/.jack-classic exists" or "environment > variable JACK_COMMUNICATION=DBUS". Or a system wide configuration: > "/etc/jackd/configuration says OSC is enabled". > > I agree with Nedko that this is not a change in the API. This is a > change in the implementation of the API. There's a huge difference as > we all know. We would run into all kinds of trouble if we would make > client software responsible for choosing the server communication > model. Hence the implementation was changed, not the API... > > > Is there anything really difficult here? Just make everyone live > happily together.
First we have to get a consensus on this "runtime choice of auto-start strategy", then we'll have to find the best way to implement it. Stephane _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
