On Monday 06 October 2003 1:40 pm, Chris Cannam wrote: > On Monday 06 Oct 2003 12:16 pm, Richard Bown wrote: > > Each MidiDevice > > _should_ have a MidiMetronome attached to it > > Hmm... is that the case at the moment?
No, it's not yet but the code I'm working on now should sort this. > Certainly the metronome _instrument_ is only created for the first > device, not for every device. (AlsaDriver.cpp:793) Yeah, I think we just forget about a Metronome Instrument (which was a hack anyway) and just use normal instruments mapped through the mechanism described. > Piss, then that _does_ mean we need to duplicate the whole tempo map > at the sequencer. (See the "quid tempo?" thread of a few days ago in > which I resisted that idea.) Otherwise the sequencer has no way to > know the current tempo (i.e. tempo at current play time) without > polling the GUI continuously during playback or having the GUI shove > tempo changes continuously at it, and either way we'd be back where > we started in terms of sequencer performance. Can't we just map tempo and timesig changes (or say "current" and "next") in the ControlBlock? That wouldn't need to be updated continuously just when a tempo or timesig changes. The sequencer only needs to know current (and potentially) the next change and when it'll happen so it can generate both metronome and sync with APIs. > If a tempo change comes in on the JACK transport during playback (is > that a possible thing?) then do we want to drop a tempo change into > the composition at the current point? I guess we have to. Well that for the moment is undefined - this is when we working in JACK slave mode (for which we have no behaviour at the moment). Until now we're only done timing master (whether it be MIDI clock or JACK) - slaving to remote tempo/timesig is something we should start to consider though (so we can slave to ardour for example). R ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Rosegarden-devel mailing list [EMAIL PROTECTED] - use the link below to unsubscribe https://lists.sourceforge.net/lists/listinfo/rosegarden-devel
