* add MSYS2 build to CI
* attempt to fix install
* misunderstood msystem
* try with base-devel
* bring back gcc
* set cmake generator
* try ninja
* add release.yml
* upload all *.zip
* fix syntax
* working release.yml
* add CHANGELOG.md
* update main.yml, delete travis.sh
* get version from tag
* remove msys2 32-bit build from main.yml
* disable translucency in vanilla strictmode
* add stub items to the menu definitions
* disable translucency percentage menu item if general_translucency == 0
* generate composites only for multipatched columns (fix Urotsuki.wad)
* Revert "generate composites only for multipatched columns (fix Urotsuki.wad)"
This reverts commit 3ce4d9057b463d3cd00d49f4918afdcbceb4f20a.
* check if the lump can be a Doom patch
* print warning to stdout
* print to stderr, function name istead of warning
* rearrange and change some settings
* delete default_skill menu item, restore saving last selected skill
* move default_complevel back to General menu
* enable fullscreen by default
* fix disable number of message lines
* restore default_skill menu item, set skill to 4 (UV) by default
* ad support for SMMU swirling flats
* formatting glitch
* shorten sequence length to 256
* create distorted flat lookup table on demand
* make init function static
* add config key and menu switch
* replace win_fopen.c with wrappers from Chocolate Doom
* always include direct.h on Windows
* add missing header files, fix gcc warning
* move wrappers to m_io.c, formatting
* fix cmakedefine
* always include string.h
* use only strcasecmp and strncasecmp
* add missing thermo.h
* include config.h
* move str(n)casecmp define to doomtype.h
* add support for -shotdir parameter, fall back to save dir
* use exe name and support up to 9999 screenshots
* set screenshotdir to -shotdir parm or fall back to -save parm or current dir
* cosmetics
* entirely rip out PCX and BMP screenshots
* unconditionally create -save and -shotdir directories
* remove obsolete variable
* fix rendering of the infinite ocean effect
First, this requires changing the type for variables used to store
angle arithmetics in R_ScaleFromGlobalAngle to angle_t.
Second, this requires checking if generated seg angles and those
stored in the WAD differ "too much" and revert back to the stored
value then.
* crispy implementation was earlier