Commit Graph

6 Commits

Author SHA1 Message Date
David Robillard
bbf4175713 Another not-quite-there-but-better commit.
Brought plugin automation into the fold of new automation system.
Fixed plugin automation, broke panner automation :] (pending Panner work).
Made AutomationController better at automatically following it's controller value (mimic what gain meter does).
Fixed some visible automation track bugs (but still broken WRT serialization).


git-svn-id: svn://localhost/ardour2/trunk@2092 d708f5d6-7413-0410-9779-e7cbd77b26cf
2007-06-30 18:41:50 +00:00
David Robillard
eb296b2c95 Reduce overhead of multi-type-ness (last Summer's SoC):
Use uint32_t instead of size_t counts (halves size of ChanCount on 64-bit).
	Shift DataType values down to eliminate subtraction every index of a ChanCount or *Set.

Allow using DataType directly as an array index (prettier/terser).
Fix some mixed spaces/tabs in file comment headers.


git-svn-id: svn://localhost/ardour2/trunk@2082 d708f5d6-7413-0410-9779-e7cbd77b26cf
2007-06-29 06:58:07 +00:00
Paul Davis
a8da89d745 optimize some performance bottlenecks; remove jack_nframes_t that crept back into the code
git-svn-id: svn://localhost/ardour2/branches/midi@1779 d708f5d6-7413-0410-9779-e7cbd77b26cf
2007-05-03 23:28:57 +00:00
David Robillard
bd1220a46d Fixes for IO port adding/removing
Working audio sends/port inserts
Send gain, panning
MIDI sends working (maybe port inserts too?)
Buffer/Port fixes (related to silence)
Metering bug fixes


git-svn-id: svn://localhost/ardour2/branches/midi@883 d708f5d6-7413-0410-9779-e7cbd77b26cf
2006-09-01 07:38:55 +00:00
David Robillard
30ab1fd615 Towards MIDI:
- Converted vector<Sample*> to BufferList and numerous counts from int to ChanCount (and related changes)
- Added fancy type-generic iterators to BufferList, PortIterator (see IO::collect_input for a good example of the idea - the same code will work to read all input (of various types in a single IO, eg instruments) without modification no matter how many types we add)
- Fixed comparison operator bugs with ChanCount (screwed up metering among other things)
- Moved peak metering into it's own object, and moved most of the pan related code out of IO to panner (still a touch more to be done here for MIDI playback)

Not directly MIDI related fixes for problems in trunk:

- Fixed varispeed gain/pan automation to work properly (was reading the wrong range of automation data, probably causing nasty clicks?)
- Fixed crash on varispeed looping (possibly only a 64-bit problem).  It still doesn't work, but at least it doesn't die

Quite a few things broken, and the new classes are pretty filthy still, but I think the direction is a lot better than all my previous plans...


git-svn-id: svn://localhost/ardour2/branches/midi@795 d708f5d6-7413-0410-9779-e7cbd77b26cf
2006-08-12 08:20:24 +00:00
David Robillard
cbdf686e39 - Replaced integer port counts (and input/output maximum/minimum) with ChanCount, which can count multiple types and does the reasonable thing for all comparison operators
- Removed the fader/meters from MIDI mixer strips, at least until they do something
- Made the Add Route dialog refuse to create MIDI busses, Spifftacular warning dialog and all

Changes a bit more widespread than I was hoping, but worked out really well - lots of code will continue to work fine even when multi-typed (eg instrument) IOs come around, just ignoring the types it doesn't care about.  Most all changes related to counts are little search/replace deals, logic doesn't need to change.  Hopefully SVN can handle (automatic) merging with the other SoC projects if the buffer change goes as well.

Next step: do for buffers what the last two commits did for ports.


git-svn-id: svn://localhost/ardour2/branches/midi@787 d708f5d6-7413-0410-9779-e7cbd77b26cf
2006-08-11 07:15:30 +00:00