On 07/03/2011 03:01 PM, Paul Davis wrote: > On Sun, Jul 3, 2011 at 8:51 AM, Robin Gareus <[email protected]> wrote: > >> Every program (on un*x) can be "quit w/o save" without requiring changes >> to the program: just send it a `kill -9` :-)) > > if (request.type == QuitWithoutSave) { > kill (getpid(), SIGKILL); > } > > oops, you didn't reply to the session manager, which wants to know how > to restart you. > > ok, so we leave it to the SM to kill you ... but using SIGKILL for > this is about as bad as LADISH using SIGUSR1 for its own purposes. > which is to say - it works, but its Not Unix :)
I should have known that you already considered this option. >> Alas, jackd may or may not bark (xrun) if a program is killed without >> properly closing its jack connection; but that's acceptable in that case >> (closing the session anyway). > > I think that the SM should tell JACK to stop calling process(), somehow. Well, sometimes one does not to see the wood for the trees.. Is "quit w/o save" really needed? IMHO a smarter way would be to "auto-save" the session every e.g. 5 mins and add keep those as revisions in a repository. JACK does not require additional code, all functionality would be done by the session-management tool. A "quit w/o save" would just "quit with save" and then roll-back to the last saved session state or snapshot (if any). 2c, robin _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/listinfo/linux-audio-dev
