The benefit to this is fairly small, but it does mean a little less work needs
to be done in the build scripts for cross compiling. The C version wasn't
especially concise so it was not obviously better in any way.
The reason for the previous delay was to allow OpenAL Soft to fade the sound to
silence and avoid a sudden amplitude change, which can cause a "click" or "pop"
noise. However, with OpenAL Soft 1.20, this is a built-in feature of the mixer
so the extra complexity is unnecessary.
CheckReturn() must be called before emitting code, otherwise it will always return false for switch statements
Redundant jump instruction added because of that will point right after the end of function's code when if statement is the last one in the given function
This wasn't a problem for VM because bytecode in question was unreachable
JIT compiler tries to generate native code for a bogus jump destination, and this leads to out-of-bounds read from labels array
https://forum.zdoom.org/viewtopic.php?t=67149
Use '-dumpdisasm something' to output functions with 'something' in their printable names
Open and close output file only once when processing code generated by Dehacked, and calculate its sizes as well
This may look as a workaround, it is not.
nosave cvars are meant to be used as a storage for statistical data that doesn't
affect gameplay. That's why this data isn't saved to the savefile in the first
place. Therefore, there is no point in sending this data over network. It would
have no meaning on machines other than local.
At the moment, we assume that dirent struct has d_type member, and DT_DIR is defined
This is true for supported versions of macOS, and Linux with glibc
https://forum.zdoom.org/viewtopic.php?t=66945
The old method using a single string with a backslash as separator is unable to handle anything with actual backslashes in the data.
It now uses a JSON object with each CVAR being a separate key.
# Conflicts:
# src/version.h
Both now use the Unicode capable I_Find* interface instead of duplicating all its functionality in multiple platform dependent incarnations.
# Conflicts:
# src/cmdlib.cpp
# src/gamedata/textures/image.cpp
# src/resourcefiles/file_directory.cpp
# Conflicts:
# src/resourcefiles/file_directory.cpp
(The Raven license got removed from sc_man after checking that no Raven code is still present here, this file got a nearly complete overhaul over all those years.)
# Conflicts:
# src/textures/texturemanager.cpp
There were two booleans that could be merged into the flag word.
This also fixes a bug with CHAN_NOSTOP not working for local sounds because it checked the wrong sound source for the playing sound.
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
* Implement renderstyle and transparent actor sprite rendering in automap
* Add "am_advspriterender" CVAR and add invisible actor checks.
* Rename cvar to `am_thingrenderstyles`
# Conflicts:
# src/am_map.cpp
- made the vid_scaleto____ commands less hacky - after finding out I could route the calls through screen->, found the correct screen-> commands, and do scaling based on the real screen dimensions
For example, SW1SKULL and SW2SKULL switches are animated in TNT: Evilution
Their frames are defined in ANIMATED lump which is old BOOM binary format
Textures other than base were not cached because the corresponding switch definitions (in ANIMDEFS lump) have one frame only and BOOM style animations were not taken into account
https://forum.zdoom.org/viewtopic.php?t=66652
# Conflicts:
# src/p_setup.cpp
Actual boost heavily depends on platform's memory architecture, made it a bit more cache friendly in general
# Conflicts:
# src/textures/hires/hqresize.cpp
Notable changes:
* IsSourcePlayingSomething had undefined behavior when checking unattached and unpositioned sounds.
* loading Blood RAW sounds with different sample rates but the same backing lump may not use the same hardware sound buffer.
* when playing an unpositioned sound the attenuation is irrelevant and must be ignored. This resulted in a 3D sound being started which was mostly inaudible due to lack of valid origin.
- added Marisa Kirisame's CHAN_OVERLAP flag.
- exported S_IsActorPlayingSomething to ZScript.
The sound API change required deprecating A_PlaySound and S_Sound. There are now new variants S_StartSound and A_StartSound which have two distinct parameters for channel and flags.
# Conflicts:
# src/bbannouncer.cpp
# src/fragglescript/t_func.cpp
# src/g_shared/a_lightning.cpp
# src/p_effect.cpp
# src/p_mobj.cpp
# src/p_switch.cpp
# src/playsim/p_spec.cpp
# src/sound/s_doomsound.cpp
# src/sound/s_doomsound.h
# wadsrc/static/zscript/base.zs
# Conflicts:
# src/intermission/intermission.cpp
# src/sound/s_doomsound.cpp
Also simplified the sound init decision making. With FMod gone there is no reason to be pedantic here. Even the check of snd_backend for the Null device could be omitted here, its only realistic use is '-nosound'.
# Conflicts:
# src/sound/s_environment.cpp
The game independent part of the code has been mostly isolated.
# Conflicts:
# src/sound/s_doomsound.cpp
# src/sound/s_sound.cpp
# src/sound/s_sound.h
# Conflicts:
# src/sound/s_sound.cpp
Many of the simple wrappers have been moved to a separate file and the sound source handling has been abstracted.
This is only the first phase, the work is not complete yet.
Also changed the license of the sound code to BSD after verifying that this code bears no similarity to id's original sound code anymore, save for a few function names (which are due to be refactored out anyway.)
# Conflicts:
# src/sound/s_sound.cpp
# src/sound/s_sound.h
src/posix/cocoa/i_video.mm:68:31: error: property 'delegate' not found on object of type 'id'
src/posix/cocoa/st_console.mm:464:37: error: property 'delegate' not found on object of type 'id'
Do not access various global objects at early stage of initialization and late stage of shutdown
# Conflicts:
# src/posix/sdl/i_main.cpp
# Conflicts:
# src/posix/sdl/i_main.cpp
src/posix/cocoa/i_main.mm:152:2: error: use of undeclared identifier 'ShutdownJoysticks'
src/posix/sdl/i_system.cpp:128:3: error: ‘I_FatalError_Gtk’ was not declared in this scope
src/posix/sdl/st_start.cpp:329:8: error: ‘CExitEvent’ was not declared in this scope
The main exits are initiated from code that cannot filter this back to D_DoomMain easily so the exception is the only way to get there.
The 3 main points of exit are:
* quit/exit CCMD
* quitting the menu through ST_Endoom
* receiving a quit message on the main window.
# Conflicts:
# src/posix/cocoa/st_start.mm
These ones were particularly bad examples of misusing the exit handlers by temporarily installing one themselves and then calling exit to clean stuff up.
Now they just return an error code to D_DoomMain to perform a regular exit.
The main catch is now in D_DoomMain, only calling platform specific functions to handle the output for the error.
As a nice side effect, -norun can now be done without an exception, just by exiting D_DoomMain with a special exit code.
# Conflicts:
# src/win32/i_main.cpp
DYN_SNDFILE=NO and DYN_MPG123=NO were ignored while DYN_FLUIDSYNTH=NO broke compilation
These options should be applied to ZMusic target instead of the main executable
As a bonus, it's now possible to build GZDoom without FluidSynth
Since this made heavy use of Windows type, those were duplicated to avoid rewriting the entire interface.
This split at least reduces the amount of code needed to refactor for making the screens work on other platforms than Windows.
# Conflicts:
# src/CMakeLists.txt
Now there is only one single entry point for both, instead of previously 2 entry and 4 exit points.
This also eliminates the explicit shutdown of ZMusic. Timidity++'s two buffers have been put in containers that self-destruct on shutdown and calling dumb_exit is not necessary because the only feature requiring it is not used by any code in the music library.
Instead of trying a homegrown way to avoid recursive exceptions, let's do it with the defined procedure C++ has for this case: call std::terminate.
This allowed removing some old hackery inherited from Boom and will now hopefully allow sanitizing the exit procedure to the point that it can be done without depending on exit handlers.
# Conflicts:
# src/files_decompress.cpp
# Conflicts:
# src/d_main.cpp
The only difference here were the size values on Windows but for this initial call they have been useless for a long time. When this code was written the console buffer still had a fixed width that needed to be set before adding any text.
FMod had MP3/Ogg playback integrated right into itself, and the OpenAL backend tried to replicate this functionality.
This functionality, however, has been removed over two years ago when FMod started breaking things more and more, it was only this backing implementation that was left in.
This was the last player class.
This code was also cleaned up for non-Windows systems where CD Audio is not implemented.
Instead of providing an empty implementation, all related code is now explicitly deactivated.
# Conflicts:
# src/CMakeLists.txt
It is now being handled by the controlling code.
While of no benefit for GZDoom itself, this finally allows to separate the entire music code into a separate, engine independent project that merely provides streamed music data when not playing on a hardware device (WinMM Midi or CD Audio.)
The tight coupling of the music code with the sound backend made this nearly impossible before
The way CDDA was treated as an afterthought made handling of stream songs somewhat problematic, because the state could be unclear.
CDDA is an easily identifiable format so it should be tested first.
Since this gets used by both the sound backend and the music code it needs to be in a place accessible to both.
# Conflicts:
# src/CMakeLists.txt
# Conflicts:
# libraries/zmusic/decoder/mpg123_decoder.cpp
# libraries/zmusic/decoder/sndfile_decoder.cpp
# src/CMakeLists.txt
This was getting a bit unwieldy. The include path setup is not perfect yet, that's work for later.
(It's about time we're getting C++20 with modules so that this include path madness can be put to an end.)
Also made the decode_vorbis function in DUMB a function pointer so that the library does not depend on high level code and can just ignore the vorbis case if no supported.
Currently all it contains are the MIDI sources and the MIDI devices, the rest needs to be reworked first.
# Conflicts:
# libraries/zmusic/i_module.cpp
# libraries/zmusic/i_module.h
# src/CMakeLists.txt
# src/i_module.cpp
# src/i_module.h
# src/sound/music/midi_cvars.cpp
# src/utility/i_module.cpp
# src/utility/i_module.h
# Conflicts:
# src/CMakeLists.txt
# src/sound/musicformats/music_opl.cpp
This isn't the final location but this means that the device is merely a data provider, with the sole exception of the Win32 MIDI device whose unwieldy usage requirements unfortunately dictate much of the needed interface here.
