Fury: fix screen return-to-center when triggering save checkpoints, hopefully for good this time. :p

Adds MODE_NEWGAME to player .gm

Uses minicoro, https://github.com/edubart/minicoro

This fixes a crash when restarting the sound system twice in a row with the OPL3 driver selected but an .ogg (or .flac or any other format) is playing instead, and makes a couple of things FORCE_INLINE in the process.

audiolib: move libasync task to VoiceNode struct, explicitly wait on task completion before attempting to service a voice with MV_BeginService()

This fixes crashes when attempting to service an .ogg voice immediately after starting playback, before the decoder has actually finished initializing.

engine: Add parameter to hictinting_applypixcolor to allow disabling r/b channel swapping. Improved readability of hictinting_applypixcolor as well.

Duke3d: add scroll wheel support back into file selection menus, add PgUp/PgDn and scroll wheel support to the little submenus used for selecting screen resolution and configuring the framerate limit etc

I'm not sure if this is going to end up being the correct thing to do here, but it's what we shipped in the Fury 2.0 binary.

Duke3d: reformat information displayed within the save and load menus, add episode name and player health

I'm not very happy with the list of values for the framerate limiter. All methods of presenting these values to the end user seem to be terrible: linear slider bars are cumbersome and contain many unnecessary values, and simply presenting a field for a value to be entered into requires the end user know what they're doing. Ugh.

TIL vsnprintf and _vsnprintf are not, in fact, the same thing.

engine: remove r_vbos cvar (useless for a while now), modify voxmodel to store data converted to tris instead of quads that are converted during the VBO building pass

This is going to need a little work to get going again on the couple of configurations that don't define USE_GLEXT (GEKKO and EDUKE32_GLES). Should USE_GLEXT even be a thing anymore?

engine: adjust polymost z-fighting mitigation, add two new cvars: r_spritedepth and r_spritedepthmul

r_spritedepth is the base multiplier for a sprite's position in the depth buffer and r_spritedepthmul controls the multiplier for a unique offset value assigned to each sprite drawn.

engine: mask angle precision related geometry intersection by calculating higher precision angle for intersecting sprite planes from intersecting wall's normal

Dino Bollinger authored Nov 06, 2021 and Richard Gobeille committed Nov 10, 2021

Engine: Fix updatesectorz() not checking z height in fallback, and add fallback optimization to updatesectorexclude()

Dino Bollinger authored Sep 18, 2021 and Richard Gobeille committed Nov 02, 2021

Usage:
```
getngcflags <var> <idx1> [<idx2>...]
setngcflags <const/var> <idx1> [<idx2>...]
```

Gets/sets the flags for the newgamechoices entries at the specified
indices. The index list is variable-sized, up to a maximum of 3.

This can be used to enable/disable the MEF_Hidden and MEF_Locked state.
Supersedes the userdef struct members:

`m_newgamecustomopen` and `m_newgamecustomsubopen`.

Dino Bollinger authored Sep 18, 2021 and Richard Gobeille committed Nov 02, 2021

Also permit usercontent flag on subentries.

audiolib: offload initial ov_open_callbacks() etc calls for vorbis into worker threads to avoid VM stalls when playing tons of .oggs from script

engine: update libdivide to b322221677351ebb11f0a42fe9a9a2794da5bfe5, add libdivide_config.h

Richard Gobeille authored Oct 28, 2021 and Richard Gobeille committed Oct 31, 2021

The warning says you can silence it by surrounding sizeof(int32_t) with extra parentheses, but it doesn't work.

Richard Gobeille authored Oct 19, 2021 and Richard Gobeille committed Oct 31, 2021

editor: remove call to calc_sector_reachability() from yax_updategrays() in favor of calling it explicitly when needed instead

Richard Gobeille authored Oct 20, 2021 and Richard Gobeille committed Oct 31, 2021

engine: print commit hash from SDL_GetRevision() to log at startup when using SDL dev builds (odd-numbered patch level)

This is admittedly terrible, but it shouldn't break unless the SDL_GetRevision() output format changes, in which case it just won't print the string anymore.