It seems 64bit builds for some reason have problems with VST Sync feature on,
workaround seems to be converting VST sync patch from double to floats,
which does work both with 32 and 64bit builds. Double precision
seems to produce odd numbers with 64bit build. (tested on VirtualBox Linux
Mint 14.1 64 bit OS)
(cherry picked from commit 011f87e6e60cccd16f3783e9c4885e03d95c1e56)
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Folowing change should ensure VST effect is opened just once, once
loaded from project file.
Double opening seems to trace back to commit 184ddc4d1c
from 2006, when this VST effect save / load feature was introduced as new.
Anyway VST effect parameters seems to load corectly in
VstEffectControls::loadSettings, from project file even without double
VST effect opening.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
This patch will set plugin name (information which is not stored with lmms
project file) according plugin file, without *.dll, when is LMMS project
loaded from the file.
Future verion could maybe use PluginBrowser or EffectSellectDialog for the
same.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
This patch should bring VST to host synchronization for LMMS.
(e.g. for plugins like dBlue Glitch, TAL Filters).
Synchronization is done via shared memory, missing song time
positions are reccalculated and added to PPQ position sync values
(SHM - common input interface for sync of all VST plugins)
Master channel keeps freezing, when VST effects are placed directly
into Master channels FxChain, on slower computers.
Provided solution prolongs VST initialisation phase, by waiting on two extra messages,
which usually follows initialisation and are possibly not yet implemented well.
IdSampleRateInformation,
IdBufferSizeInformation
This seems to prevent Master channel freezes e.g. on VST effects like Tiny-Q.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Various fixes improvements:
+ Added support for VST parameters control for windows.
+ New `Close` button for VST parameter controls.
+ Faster GUI for all instruments, effects-loading, not only VSTs,
and both one-instrument track window mode and normal window
mode should be supported.
+ Better integration for VST GUIs on Linux, e.g. plugin window
should not stay always on top of other windows.
+ VST GUI overlook should remain same with different wine setups
( except for whole virtual desktops emulations ).
+ VST effect control window merged with VST effect editor window
should be more easier to control.
+ Little corections at effectviews model updates of instrument
tracks effect chains.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
When sending MIDI events to the ZynAddSubFX engine do not statically
send them on channel 0 but on the MIDI channel which is set as output
MIDI channel of the instrument track. This adds some flexibility when
it comes to multipart ZynAddSubFX presets which are connected to
different MIDI channels.
Window title for VSTi parameter controlling window should be set according to
actual track name, not from VST plugin name as its now, it is hard to get to know
which window belongs to where when same plugin is e.g. opened several times, so
with the same name. Now this is handled in paint event, but TBD whenever is
that track name changed. tbd, temporary solution.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
When you double-click on some knob to change its VST parameter value
manually, new dialog window now has same title as what was that knobs
name. (instead of "lmms" title string)
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Filter to display only automated / all knobs (new button) on LMMS VST parametr control window.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
After project save/load unsaved VST control parameters in LMMS VST control wrapper are not set to zero now, but reloaded according plugins saved state.
No need to sync values manualy again, after project save, load.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Only not automated values are synced from VST plugin, no need to sync already automated values.
Synced values are now not trackable with undo / redo, before all changes were tracked individualy,
but user could lose ability to undo easily changes before sync buton was pressed.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Fixed destuctors for various types of VST automation panel removal.
Automated signals correct disconection on panel removal.
When you load new VST plugin in the place of old one, while automation
is already connected and active there, this should not result in LMMS crash
or reuse of that old automation connections and automation window destructor
should not leave zombie VST plugins on application exit.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
src/core/RemotePlugin.cpp @ RemotePlugin::process
Above function should be now more thread safe, but functionality remains.
This prevent lmms locks, when automation is connected to lmms VST plugin wrappers controler / knob on Linux.
On win32 build whenever is lmms wrapper parameter controler / knob amended.
plugins/vst_base/VstPlugin.cpp @ VstPlugin::setParam
plugins/vst_base/RemoteVstPlugin.cpp @ RemoteVstPlugin::processMessage
In above functions we dont wait for message confirmation when parameter setter function finish, and dont send such confirmations.
This workaround prevent locks on Linux, whenever there is automation connected and if you try to move with VST plugin
(not via lmms plugins wrapper, but with VSTs internal window) than it freezes, on win32 build it will freeze whenever you connect
automation and than if you move your mouse above this VST plugin window.
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Not all plugins save current program in their chunk (or do not restore
it properly). We therefore have to save and restore the current program
manually.
Closes#3581879.
The code for managing programs/presets of RemotePlugin instances was
very confusing, mainly within the VstPlugin and RemoteVstPlugin class.
I therefore started to reorganize and rewrite functions.
We have too many components in LMMS relying on FFTW3. Building LMMS
without them cripples LMMS' functionality substantially, so simply
make fftw3 a requirement.
Closes#3495736.
This patch includes:
* sampleBuffer::visualise(): add possibility to specified a range to visualize instead of the whole sample
* add sampleBuffer::sampleRate() and sampleBuffer::sampleLength() getters
* definition of AudioFileProcessorWaveView and AudioFileProcessorWaveView::knob classes for AudioFileProcessor plugin
* knob::getValue() specified “virtual” to allow redefinition in child class AudioFileProcessorWaveView::knob
* delete audioFileKnob class (made obsolete by AudioFileProcessorWaveView::knob)
* add audioFileProcessor::isPlaying() signal, which is emitted in audioFileProcessor::playNote
* change type of AudioFileProcessorView::m_startKnob and AudioFileProcessorView::m_endKnob (AudioFileProcessorWaveView::knob instead of audioFileKnob)
* replace AudioFileProcessorView::m_graph (QPixmap) by AudioFileProcessorView::m_waveView (AudioFileProcessorWaveView)
Signed-off-by: Tobias Doerffel <tobias.doerffel@gmail.com>
Upstream commit 460642bfae534679b4094e6887714828c90d8172 removes LADSPA
support from CALF which definitely is not desired for LMMS. Therefore
simply revert it in our sources.
When building LMMS with recent stable versions of GCC it generates
wrong code for the implicit REALTYPE specialization of the template
method stringFrom<T>() (problem in std::stringstream implementation).
By writing a custom template specialization which uses sprintf() the
bug can be circumvented.
