diff options
author | Tim E. Real <termtech@rogers.com> | 2011-02-09 22:37:27 +0000 |
---|---|---|
committer | Tim E. Real <termtech@rogers.com> | 2011-02-09 22:37:27 +0000 |
commit | d7f4820023196bfc7d40ef5fa134bc46a6e55849 (patch) | |
tree | fa00ee04c060c2ece1122f9d3fee382b0d5f043c /muse2/ChangeLog | |
parent | e11f572d97bb761d848fad41991d0ed747bad111 (diff) |
Mini MIDI overhaul. Please see ChangeLog.
Diffstat (limited to 'muse2/ChangeLog')
-rw-r--r-- | muse2/ChangeLog | 20 |
1 files changed, 20 insertions, 0 deletions
diff --git a/muse2/ChangeLog b/muse2/ChangeLog index 081649b8..9928262c 100644 --- a/muse2/ChangeLog +++ b/muse2/ChangeLog @@ -1,5 +1,25 @@ 09.02.2011: - fixed regression with Bounce to File and Bounce to Track (rj) + [[[ By Tim. Marked as p4.0.15 ... + - Removed MidiDevice::nextPlayEvent, and instead of erasing 'already played events' at top of + Audio::processMidi(), let each device erase played events immediately after playing them. + Still a bit uneasy with this one, but nextPlayEvent was causing some problems. + This required forced event erasure, if the track is off, in SynthI::preProcessAlways(), + which may need more tweaking for other non-processed circumstances. + - Added MidiDevice::putEventWithRetry() which will wait and retry sending an event. + So far it is only used in MidiPort::setMidiDevice(), which runs in the GUI thread where + it can wait a bit. + - Increased playback MidiFifo size from 512 to 2100, to accommodate large block sends, + such as in MidiInstrument::reset() which sends 2048 events at once. + Created new recording MidiRecFifo with reduced size of 160 instead of previous 512. + Replaced SynthI::putFifo and MidiJackDevice::eventFifo with unified MidiDevice::eventFifo. + TODO: Allow ALSA devices to also make use of it, in case events cannot be delivered. + - Added a timestamp to OscControlFifo events. To be used for better, faster OSC processing - TODO. + - Fixed DSSI processing timing in DssiSynthIF::getData(). Should be much better now, tested OK. + - Fixed Jack Midi 'lost event' messages due to full buffers. Now it waits until next cycle if buffer is full. + Slight mod: SysEx events forced to be lost, since it is most likely that the sysex is too big and Jack Midi + could never allocate enough space for it, meaning the event would linger around forever attempting to + be sent, tying up other events. ]]] 08.02.2011: - made muse compilable under gcc-4.6. ptrdiff_t type requires inclusion of cstddef. (Orcan) 31.01.2011: |