Significant internal changes to sound streams:
Abstracted buffers of sound data into an internal stream_buffer class, with helper classes read_stream_view and write_stream_view which offer readable/writable "views" into the buffers
Internal sound calculations are all done using stream_buffer::sample_t, which is a 32-bit float; existing callbacks are supported through an adapter that converts to/from signed 32-bit integers
Improved behavior of dynamic stream sample rate changes to resample a short runway of data to preserve continuity across transitions
Created a new stream update callback which passes a std::vector of read_stream_views for inputs, and a std::vector of write_stream_views for outputs
Updated core mixer and speaker devices to the new stream update callback
Updated the following sound cores to the new stream update callback: ay8910, dac, k054539, msm5205, namco, netlist, okim6295, pokey, samples, sn76496, sp0250, tms5220, tms57002, upd7759, vgm_visualizer, volt_reg
Changed existing stream update callback to make inputs explicitly const and the output pointers const as well, since they are re-used across calls; fixed several engines that violated this rule
Sound_manager::stream_alloc can no longer automatically connect to a device's sound_stream_update callback; instead, the stream_alloc() on the sound_device_interface should be called; updated many violators of this rule
Streams can be created with SAMPLE_RATE_OUTPUT_ADAPTIVE, which dynamically tracks the sample rate of its first downstream output, or with SAMPLE_RATE_INPUT_ADAPTIVE, which tracks the sample rate of its first input
Changed resampling to be a separate sound_stream that is invoked as needed, opening the path for selectable resampling implementations
Added a flags parameter to the new stream allocation method that allows you to specify a that input streams should not be resampled
Exposed stream_input and stream_output classes directly, simplifying access to user gains and stream names
Added a simple dynamic compressor to sound_manager to provide nicer results when overdriven sound happens; compression does not affect speaker_report results
Improved verbose speaker_report to print a graph of peaks over time
More aggressive debugging enabled for now even in release builds (should be disabled prior to next release) via SOUND_DEBUG define in sound.h; report any assertions for fixing
----------------------------------------------------------
The Last Ninja [4am, Firehawke]
Centauri Alliance [4am, Firehawke]
Apple Galaxian (StarCraft) [4am, Firehawke]
Apple World [4am, Firehawke]
* Removed device and macro header files.
* All of those can be generated automatically so going forward there is
no need for these any longer.
* Introduced the modules concept. Modules are netlists for which
automatic lib entries are generated.
* Going forward you just store them in macro/modules and they will be
automatically registered as device elements.
* You need to do a "make generated" is src/lib/netlist/build
* Some_device.cpp still needs to be added to netlist.lua
* Added documentation on how to add devices to netlist.
* Please refer to adding_devices.md for more information.
