Fixed an incorrect timer processing when using a real-time interface.
This bug does directly affect the case of ZMusic which does use of RealTime API of both libraries.
Different volume models were means louder or quiter sounding of the rest of notes in the song. And to avoid the mess between volume models, let's use different gain factor for each volume model?
After a small set of tests, it's fine to use the "AUTO" volume model by default.
Every embedded bank and every WOPL file has a setting of a volume model that matches to the behavior of the original OPL2/3 driver of each volume model.
## 1.5.0 2020-09-28
* Drum note length expanding is now supported in real-time mode (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Added support for OPNA chip with Neko Project II Kai YM2602 emulator usage (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Added VGM file dumper which allows to output OPN2 commands into VGM file. (A new MIDI to VGM tool is now created with basing on libOPNMIDI)
* Fixed an incorrect work of CC-121 (See https://github.com/Wohlstand/libADLMIDI/issues/227 for details)
* Internality has been refactored and improved
## 1.5.0 2020-09-28
* Drum note length expanding is now supported in real-time mode (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Channels manager has been improved (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Nuked OPL3 1.8 emulator got some optimizations ported from 1.7 where they are was applied previously (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Reworked rhythm-mode percussions system, WOPL banks with rhythm-mode percussions
* Added Public Domain Opal OPL3 emulator made by Reality (a team who originally made the Reality Adlib Tracker) (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Added LGPL licensed JavaOPL3 emulator made by Robson Cozendey in Java and later rewritten into C++ for GZDoom (Thanks to [Jean Pierre Cimalando](https://github.com/jpcima) for a work!)
* Fully rewritten an embedded bank database format, embedded banks now supports a wider set (more than 127:127 instruments in one bank)
* Improved accuracy of the DMX volume model, include the buggy AM interpretation
* Improved accuracy of Apogee volume model, include the bug of AM instruments
* Improved accuracy of the Win9X volume model
* Removed C++ extras. C++-bounded instruments tester is useless since a real-time MIDI API can completely replace it
* Added AIL volume model
* Added Generic FM variant of Win9X volume model
* Fixed an incorrect work of CC-121 (See https://github.com/Wohlstand/libADLMIDI/issues/227 for details)
* Added HMI volume model (Thanks to [Alexey Khokholov](https://github.com/nukeykt) for help with research!)
* Added frequency models, assigned to every volume model: AIL, HMI, DMX, Apogee, 9X, and the Generic formula
# Conflicts:
# libraries/adlmidi/adldata.cpp
# libraries/adlmidi/adldata.hh
# libraries/adlmidi/adlmidi.hpp
Let's hope nothing got broken on the Linux side, the recent submission's code was not usable.
# Conflicts:
# libraries/zmusic/zmusic/configuration.cpp
# src/sound/music/i_music.cpp
# src/sound/music/music_midi_base.cpp
There is a high chance of them getting called within the stream servicing function which cannot deal with abnormal conditions, so the only choice here is performing a hard abort.
If ZMusic is to act like an external library it may not call delete on external objects because there is no guarantee that they use the same allocator. Deletion must be done as a virtual function to ensure that the correct operator delete gets called, which, unlike the actual destructor is not virtual itself.
Most of the synchronization was too deep in the implementation so that it did not guard everything it needed.
Now each song has precisely one mutex which must be locked for all access to its internals - this is done in the public ZMusic interface
FluidSynth 1.x: these functions return 1 on success and 0 otherwise
FluidSynth 2.x: these functions return FLUID_OK (0) on success and FLUID_FAILED (-1) otherwise
