work towards DeferredCommandDraw being universal, support for DeferredCommandDepthRange, rework of cirlce mesh (has a bug still), get_quad_element_buffer() now more general, as it should be with gl_any

CMakeLists.txt

@@ -9,11 +9,11 @@ if(NOT EMSCRIPTEN)
 endif()
 
 # CMake actually has no default configuration and it's toolchain file dependent, set debug if not specified
-if(NOT CMAKE_BUILD_TYPE)
+if(NOT DEFINED CMAKE_BUILD_TYPE)
         set(CMAKE_BUILD_TYPE Debug)
 endif()
 
-if(NOT TWN_SANITIZE AND CMAKE_BUILD_TYPE MATCHES Debug)
+if(NOT DEFINED TWN_SANITIZE AND CMAKE_BUILD_TYPE MATCHES Debug)
         set(TWN_SANITIZE ON)
 endif()
 
@@ -24,6 +24,7 @@ set(TWN_ROOT_DIR ${CMAKE_CURRENT_SOURCE_DIR} CACHE INTERNAL "")
 
 # feature configuration, set them with -DFEATURE=ON/OFF in cli
 option(TWN_FEATURE_DYNLIB_GAME  "Enable dynamic library loading support" ON)
+option(TWN_USE_AMALGAM  "Enable use of twn_amalgam.c as a single compilation unit" ON)
 
 # todo: figure out how to compile for dynamic linking instead
 if(EMSCRIPTEN)
@@ -41,19 +42,21 @@ if(HAIKU)
 endif()
 
 # add -fPIC globally so that it's linked well
-add_compile_options($<$<BOOL:${TWN_FEATURE_DYNLIB_GAME}>:-fPIC>)
+add_compile_options($<$<BOOL:${TWN_FEATURE_DYNLIB_GAME}>:-fPIC>
+                    -fvisibility=hidden)
 
-set(PHYSFS_BUILD_SHARED FALSE)
-set(PHYSFS_DISABLE_INSTALL TRUE)
-set(PHYSFS_TARGETNAME_UNINSTALL "physfs_uninstall")
-set(PHYSFS_ARCHIVE_GRP OFF)
-set(PHYSFS_ARCHIVE_WAD OFF)
-set(PHYSFS_ARCHIVE_HOG OFF)
-set(PHYSFS_ARCHIVE_MVL OFF)
-set(PHYSFS_ARCHIVE_QPAK OFF)
-set(PHYSFS_ARCHIVE_SLB OFF)
-set(PHYSFS_ARCHIVE_ISO9660 OFF)
-set(PHYSFS_ARCHIVE_VDF OFF)
+set(PHYSFS_BUILD_SHARED FALSE CACHE INTERNAL "")
+set(PHYSFS_DISABLE_INSTALL TRUE CACHE INTERNAL "")
+set(PHYSFS_TARGETNAME_UNINSTALL "physfs_uninstall" CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_GRP OFF CACHE BOOL "")
+set(PHYSFS_ARCHIVE_WAD OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_HOG OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_MVL OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_QPAK OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_SLB OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_ISO9660 OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_VDF OFF CACHE INTERNAL "")
+set(PHYSFS_ARCHIVE_7Z OFF CACHE INTERNAL "")
 add_subdirectory(third-party/physfs ${CMAKE_CURRENT_BINARY_DIR}/third-party/physfs SYSTEM)
 add_subdirectory(third-party/libxm ${CMAKE_CURRENT_BINARY_DIR}/third-party/libxm SYSTEM)
 
@@ -88,7 +91,8 @@ set(TWN_THIRD_PARTY_SOURCE_FILES
         third-party/tomlc99/toml.c
         $<$<NOT:$<BOOL:${EMSCRIPTEN}>>:third-party/glad/src/glad.c>)
 
-set(TWN_SOURCE_FILES
+set(TWN_NONOPT_SOURCE_FILES
+        src/twn_stb.c
         src/twn_loop.c
         src/twn_main.c
         src/twn_context.c include/twn_context.h
@@ -100,11 +104,15 @@ set(TWN_SOURCE_FILES
 
         src/rendering/twn_draw.c src/rendering/twn_draw_c.h
         src/rendering/twn_sprites.c
+        src/rendering/twn_rects.c
         src/rendering/twn_text.c
         src/rendering/twn_triangles.c
         src/rendering/twn_circles.c
         src/rendering/twn_skybox.c
-        src/rendering/twn_fog.c
+        src/rendering/twn_fog.c)
+
+set(TWN_SOURCE_FILES
+        $<IF:$<BOOL:${TWN_USE_AMALGAM}>,src/twn_amalgam.c src/twn_stb.c,${TWN_NONOPT_SOURCE_FILES}>
 
         # for dynamic load based solution main is compiled in a separate target
         $<$<NOT:$<BOOL:${TWN_FEATURE_DYNLIB_GAME}>>:src/twn_main.c
@@ -113,6 +121,7 @@ set(TWN_SOURCE_FILES
         ${SYSTEM_SOURCE_FILES})
 
 list(TRANSFORM TWN_SOURCE_FILES PREPEND ${TWN_ROOT_DIR}/)
+source_group(TREE ${TWN_ROOT_DIR} FILES ${TWN_NONOPT_SOURCE_FILES})
 
 add_library(twn_third_parties STATIC ${TWN_THIRD_PARTY_SOURCE_FILES})
 
@@ -123,8 +132,6 @@ else()
         add_library(${TWN_TARGET} STATIC ${TWN_SOURCE_FILES} ${twn_third_parties})
 endif()
 
-source_group(TREE ${TWN_ROOT_DIR} FILES ${TWN_SOURCE_FILES})
-
 set_target_properties(${TWN_TARGET} PROPERTIES
                                     C_STANDARD 11
                                     C_STANDARD_REQUIRED ON
@@ -133,7 +140,8 @@ set_target_properties(${TWN_TARGET} PROPERTIES
 # precompile commonly used not-so-small headers
 target_precompile_headers(${TWN_TARGET} PRIVATE
         $<$<NOT:$<BOOL:${EMSCRIPTEN}>>:third-party/glad/include/glad/glad.h>
-        third-party/stb/stb_ds.h)
+        ${SDL2_INCLUDE_DIR}/SDL.h
+        third-party/physfs/src/physfs.h)
 
 
 function(give_options_without_warnings target)
@@ -168,20 +176,19 @@ function(give_options_without_warnings target)
         target_compile_options(${target} PUBLIC
                                          ${BUILD_FLAGS}
                                          $<$<CONFIG:Release>:${BUILD_FLAGS_RELEASE}>
-                                         $<$<CONFIG:Debug>:${BUILD_FLAGS_DEBUG}>
-                                         $<$<BOOL:${LINUX}>:-Wl,-rpath,$ORIGIN/>)
+                                         $<$<CONFIG:Debug>:${BUILD_FLAGS_DEBUG}>)
 
         target_link_options(${target} PUBLIC
                                       ${BUILD_FLAGS}
                                       # -Wl,--no-undefined # TODO: use later for implementing no-libc
                                       $<$<CONFIG:Release>:${BUILD_FLAGS_RELEASE}>
                                       $<$<CONFIG:Debug>:${BUILD_FLAGS_DEBUG}>
-                                      $<$<BOOL:${LINUX}>:-Wl,-rpath,$ORIGIN/>
                                       -Bsymbolic-functions
-                                      -Wl,--hash-style=gnu)
+                                      $<$<BOOL:${LINUX}>:-Wl,--hash-style=gnu>)
 
         target_compile_definitions(${target} PRIVATE
-                                             $<$<BOOL:${TWN_FEATURE_DYNLIB_GAME}>:TWN_FEATURE_DYNLIB_GAME>)
+                                             $<$<BOOL:${TWN_FEATURE_DYNLIB_GAME}>:TWN_FEATURE_DYNLIB_GAME>
+                                             $<$<BOOL:${LINUX}>:_GNU_SOURCE>)
 endfunction()
 
 
@@ -193,7 +200,8 @@ function(give_options target)
                 -Wno-padded
                 -Wno-declaration-after-statement
                 -Wno-unsafe-buffer-usage
-                -Wno-unused-command-line-argument)
+                -Wno-unused-command-line-argument
+                -Wno-covered-switch-default)
 
         set(WARNING_FLAGS
                 -Wall
@@ -257,6 +265,8 @@ function(use_townengine target sources output_directory)
                 # launcher binary, loads game and engine shared library
                 add_executable(${target} ${TWN_ROOT_DIR}/src/twn_main.c)
 
+                target_link_options(${target} PRIVATE $<$<BOOL:${LINUX}>:-Wl,-rpath,$ORIGIN/>)
+
                 # todo: copy instead?
                 # put libtownengine.so alongside the binary
                 set_target_properties(${TWN_TARGET} PROPERTIES

apps/demos/bunnymark/game.c

@@ -13,11 +13,11 @@
 #define RIGHT_CLICK_ADD 500
 
 
-void handle_input(void)
+static void handle_input(void)
 {
     State *state = ctx.udata;
 
-    if (ctx.mouse_window_position.y <= 60)
+    if (ctx.mouse_position.y <= 60)
         return;
 
     if (input_is_action_pressed("add_a_bit"))
@@ -27,10 +27,12 @@ void handle_input(void)
             if (state->bunniesCount < MAX_BUNNIES)
             {
                 state->bunnies[state->bunniesCount].position = input_get_action_position("add_a_bit");
-                state->bunnies[state->bunniesCount].speed.x = (float)(rand() % 500 - 250) / 60.0;
-                state->bunnies[state->bunniesCount].speed.y = (float)(rand() % 500 - 250) / 60.0;
+                state->bunnies[state->bunniesCount].speed.x = (float)(rand() % 500 - 250) / 60.0f;
+                state->bunnies[state->bunniesCount].speed.y = (float)(rand() % 500 - 250) / 60.0f;
                 state->bunnies[state->bunniesCount].color =
-                    (Color){rand() % 190 + 50, rand() % 160 + 80, rand() % 140 + 100, 255};
+                    (Color){(uint8_t)(state->bunniesCount % 190 + 50),
+                            (uint8_t)((state->bunniesCount + 120) % 160 + 80),
+                            (uint8_t)((state->bunniesCount + 65) % 140 + 100), 255};
                 state->bunniesCount++;
             }
         }
@@ -43,10 +45,12 @@ void handle_input(void)
             if (state->bunniesCount < MAX_BUNNIES)
             {
                 state->bunnies[state->bunniesCount].position = input_get_action_position("add_a_lot");
-                state->bunnies[state->bunniesCount].speed.x = (float)(rand() % 500 - 250) / 60.0;
-                state->bunnies[state->bunniesCount].speed.y = (float)(rand() % 500 - 250) / 60.0;
+                state->bunnies[state->bunniesCount].speed.x = (float)(rand() % 500 - 250) / 60.0f;
+                state->bunnies[state->bunniesCount].speed.y = (float)(rand() % 500 - 250) / 60.0f;
                 state->bunnies[state->bunniesCount].color =
-                    (Color){rand() % 190 + 50, rand() % 160 + 80, rand() % 140 + 100, 255};
+                   (Color){(uint8_t)(state->bunniesCount % 190 + 50),
+                            (uint8_t)((state->bunniesCount + 120) % 160 + 80),
+                            (uint8_t)((state->bunniesCount + 65) % 140 + 100), 255};
                 state->bunniesCount++;
             }
         }
@@ -73,32 +77,29 @@ void game_tick(void)
 
     State *state = ctx.udata;
 
-    const double delta =
-        (double)(ctx.delta_time) / 1000.0; // Receiving floating point delta value (diving by 1000 based on vibe)
-
     for (int i = 0; i < state->bunniesCount; i++)
     {
         state->bunnies[i].position.x += state->bunnies[i].speed.x;
         state->bunnies[i].position.y += state->bunnies[i].speed.y;
 
-        if (((state->bunnies[i].position.x + BUNNY_W / 2) > ctx.base_draw_w) ||
-            ((state->bunnies[i].position.x + BUNNY_W / 2) < 0))
+        if (((state->bunnies[i].position.x + (float)BUNNY_W / 2) > (float)ctx.resolution.x) ||
+            ((state->bunnies[i].position.x + (float)BUNNY_W / 2) < 0))
             state->bunnies[i].speed.x *= -1;
-        if (((state->bunnies[i].position.y + BUNNY_H / 2) > ctx.base_draw_h) ||
-            ((state->bunnies[i].position.y + BUNNY_H / 2 - 60) < 0))
+        if (((state->bunnies[i].position.y + (float)BUNNY_H / 2) > (float)ctx.resolution.y) ||
+            ((state->bunnies[i].position.y + (float)BUNNY_H / 2 - 60) < 0))
             state->bunnies[i].speed.y *= -1;
     }
 
     handle_input();
 
     // Clear window with Gray color (set the background color this way)
-    draw_rectangle((Rect){0, 0, ctx.base_draw_w, ctx.base_draw_h}, GRAY);
+    draw_rectangle((Rect){0, 0, (float)ctx.resolution.x, (float)ctx.resolution.y}, GRAY);
 
     for (int i = 0; i < state->bunniesCount; i++)
     { // Draw each bunny based on their position and color, also scale accordingly
         m_sprite(m_set(path, "wabbit_alpha.png"),
-                 m_set(rect, ((Rect){.x = (int)state->bunnies[i].position.x,
-                                     .y = (int)state->bunnies[i].position.y,
+                 m_set(rect, ((Rect){.x = state->bunnies[i].position.x,
+                                     .y = state->bunnies[i].position.y,
                                      .w = BUNNY_W * SPRITE_SCALE,
                                      .h = BUNNY_H * SPRITE_SCALE})),
                  m_opt(color, (state->bunnies[i].color)), m_opt(stretch, true), );

apps/demos/bunnymark/state.h

@@ -3,7 +3,7 @@
 
 #include "twn_game_api.h"
 
-#define MAX_BUNNIES 100000 // 100K bunnies limit
+#define MAX_BUNNIES 500000 // 100K bunnies limit
 #define BUNNY_W 26
 #define BUNNY_H 37
 #define SPRITE_SCALE 1

apps/demos/platformer/data/packs/data.toml

@@ -1,3 +1,3 @@
 [[deps]]
-source = "../../common-data" # where does it come from, might be an url
+source = "../../../common-data" # where does it come from, might be an url
 name = "common-data"         # should be globally unique

apps/demos/platformer/scenes/title.c

@@ -19,15 +19,13 @@ static void title_tick(State *state) {
         return;
     }
 
-
     m_sprite("/assets/title.png", ((Rect) {
-            ((float)ctx.base_draw_w / 2) - ((float)320 / 2), 64, 320, 128 }));
-    
+            ((float)ctx.resolution.x / 2) - ((float)320 / 2), 64, 320, 128 }));
     
     /* draw the tick count as an example of dynamic text */
-    size_t text_str_len = snprintf(NULL, 0, "%lu", state->ctx->tick_count) + 1;
+    size_t text_str_len = snprintf(NULL, 0, "%lu", state->ctx->frame_number) + 1;
     char *text_str = cmalloc(text_str_len);
-    snprintf(text_str, text_str_len, "%lu", state->ctx->tick_count);
+    snprintf(text_str, text_str_len, "%lu", state->ctx->frame_number);
 
     const char *font = "fonts/kenney-pixel.ttf";
     int text_h = 32;

apps/demos/scenery/data/packs/data.toml

@@ -1,3 +1,3 @@
 [[deps]]
-source = "../../common-data" # where does it come from, might be an url
+source = "../../../common-data" # where does it come from, might be an url
 name = "common-data"         # should be globally unique

apps/demos/scenery/scenes/ingame.c

@@ -17,8 +17,8 @@ static void ingame_tick(State *state) {
 
     if (input_is_mouse_captured()) {
         const float sensitivity = 0.6f; /* TODO: put this in a better place */
-        scn->yaw += (float)ctx.mouse_relative_position.x * sensitivity;
-        scn->pitch -= (float)ctx.mouse_relative_position.y * sensitivity;
+        scn->yaw += (float)ctx.mouse_movement.x * sensitivity;
+        scn->pitch -= (float)ctx.mouse_movement.y * sensitivity;
         scn->pitch = clampf(scn->pitch, -89.0f, 89.0f);
 
         const float yaw_rad = scn->yaw * (float)DEG2RAD;
@@ -62,8 +62,8 @@ static void ingame_tick(State *state) {
 
     for (int ly = 64; ly--;) {
         for (int lx = 64; lx--;) {
-            float x = SDL_truncf(scn->cam.pos.x + 32 - lx);
-            float y = SDL_truncf(scn->cam.pos.z + 32 - ly);
+            float x = SDL_truncf(scn->cam.pos.x + 32 - (float)lx);
+            float y = SDL_truncf(scn->cam.pos.z + 32 - (float)ly);
 
             float d0 = stb_perlin_noise3((float)x       * TERRAIN_FREQUENCY, (float)y       * TERRAIN_FREQUENCY, 0, 0, 0, 0) * 3 - 6;
             float d1 = stb_perlin_noise3((float)(x + 1) * TERRAIN_FREQUENCY, (float)y       * TERRAIN_FREQUENCY, 0, 0, 0, 0) * 3 - 6;
@@ -89,7 +89,7 @@ static void ingame_tick(State *state) {
     }
 
     draw_skybox("/assets/miramar/miramar_*.tga");
-    draw_fog(0.9, 1.0, 0.05, (Color){ 140, 147, 160, 255 });
+    draw_fog(0.9f, 1.0f, 0.05f, (Color){ 140, 147, 160, 255 });
 }
 
 

apps/demos/scenery/scenes/title.c

@@ -16,15 +16,13 @@ static void title_tick(State *state) {
         return;
     }
 
-
     m_sprite("/assets/title.png", ((Rect) {
-            ((float)ctx.base_draw_w / 2) - ((float)320 / 2), 64, 320, 128 }));
-    
-    
+            ((float)ctx.resolution.x / 2) - ((float)320 / 2), 64, 320, 128 }));
+
     /* draw the tick count as an example of dynamic text */
-    size_t text_str_len = snprintf(NULL, 0, "%lu", state->ctx->tick_count) + 1;
+    size_t text_str_len = snprintf(NULL, 0, "%llu", state->ctx->frame_number) + 1;
     char *text_str = cmalloc(text_str_len);
-    snprintf(text_str, text_str_len, "%lu", state->ctx->tick_count);
+    snprintf(text_str, text_str_len, "%llu", state->ctx->frame_number);
 
     const char *font = "/fonts/kenney-pixel.ttf";
     int text_h = 32;
@@ -39,8 +37,8 @@ static void title_tick(State *state) {
         },
         (Color) { 0, 0, 0, 255 }
     );
-    draw_text(text_str, (Vec2){ 0, 0 }, text_h, (Color) { 255, 255, 255, 255 }, font);
 
+    draw_text(text_str, (Vec2){ 0, 0 }, text_h, (Color) { 255, 255, 255, 255 }, font);
     free(text_str);
 }
 

bin/build.sh

@@ -1,7 +1,12 @@
 #!/bin/env sh
 
-if [ "$1" = "web" ]; then
-    emcmake cmake -B .build-web "${@:2}" && cmake --build .build-web --parallel
-else
-    cmake -B .build "$@" && cmake --build .build --parallel
+# check whether ninja is around (you better start running)
+if [ -x "$(command -v ninja)" ]; then
+    generator="-G Ninja"
+fi
+
+if [ "$1" = "web" ]; then
+    emcmake cmake $generator -B .build-web "${@:2}" && cmake --build .build-web --parallel
+else
+    cmake $generator -B .build "$@" && cmake --build .build --parallel
 fi

bin/twn

@@ -16,7 +16,7 @@ case "$1" in
                ;;
 
     gdb      ) unset DEBUGINFOD_URLS
-               $0 build && gdb -ex run --args "$(basename $PWD)" "${@:2}"
+               $0 build && gdb --se=libgame.so -ex run --args "$(basename $PWD)" "${@:2}"
                ;;
 
     apitrace ) case "$2" in

docs/getting_and_compiling.txt

@@ -1,6 +1,7 @@
 firstly, make sure to pull git-lfs references with: :git lfs fetch origin main:
+you might have to run `git lfs install` before that
 install cmake and sdl2-dev packages in manner suitable to your distribution
-add /tools/ directory to your $PATH, for example, with :source hooks:
+add /bin/ directory to your $PATH, you can do it cleanly with :source hooks: command
 navigate to one of /apps/ folders for demos
 run `twn build` to build
 runnable apps should have the same name as app folder it is in

docs/interop.txt

@@ -4,6 +4,7 @@ for that certain steps are taken:
 * number of public api calls is kept at the minimum
 * procedure signatures can only use basic types, no aggregates, with exception of Vec/Matrix types and alike,
   with no expectation on new additions (see /include/twn_types.h)
+* optionals can be expressed via pointer passage of value primitives, with NULL expressive default
 * /include/twn_game_api.json file is hand-kept with a schema to aid automatic generation and other tooling
 
 one of main inspirations for that is opengl model

include/twn_camera.h

@@ -1,7 +1,7 @@
 #ifndef TWN_CAMERA_H
 #define TWN_CAMERA_H
 
-#include "twn_util.h"
+#include "twn_types.h"
 #include "twn_engine_api.h"
 
 /* TODO: make it cached? */

include/twn_config.h

@@ -1,25 +0,0 @@
-#ifndef TWN_CONFIG_H
-#define TWN_CONFIG_H
-
-/* TODO: consider if it's still necessary to keep these in one place */
-
-#define PACKAGE_EXTENSION "btw"
-
-#define TICKS_PER_SECOND_DEFAULT 60
-
-#define BASE_RENDER_WIDTH_DEFAULT 640
-#define BASE_RENDER_HEIGHT_DEFAULT 360
-
-#define TEXTURE_ATLAS_SIZE_DEFAULT 2048
-#define TEXTURE_ATLAS_BIT_DEPTH 32
-#define TEXTURE_ATLAS_FORMAT SDL_PIXELFORMAT_RGBA32
-
-#define KEYBIND_SLOTS_DEFAULT 3
-
-#define AUDIO_FREQUENCY 48000
-
-#define TEXT_FONT_TEXTURE_SIZE_DEFAULT 2048
-#define TEXT_FONT_OVERSAMPLING_DEFAULT 4
-#define TEXT_FONT_FILTERING_DEFAULT TEXTURE_FILTER_LINEAR
-
-#endif

include/twn_context.h

@@ -1,47 +1,46 @@
 #ifndef TWN_CONTEXT_H
 #define TWN_CONTEXT_H
 
-#include "twn_input.h"
+#include "twn_types.h"
 #include "twn_engine_api.h"
 
 #include <stdbool.h>
 #include <stdint.h>
 
 
-/* context that is valid for current frame */
-/* changes to it should not have an effect, unless specified */
-/* TODO: ensure the statement above */
+/* context that is only valid for current frame */
+/* any changes to it will be dropped, unless explicitly stated */
 typedef struct Context {
-    /* you may read from and write to these from game code */
+    /* you may read from and write to this one from game code */
+    /* it is ensured to persist between initializations */
     void *udata;
 
-    /* TODO: is it what we actually want? */
-    int64_t delta_time; /* preserves real time frame delta with no manipilation */
-    uint64_t tick_count;
+    /* which frame is it, starting from 0 at startup */
+    uint64_t frame_number;
 
-    Vec2i mouse_window_position;
-    Vec2i mouse_relative_position;
+    /* real time spent on one frame (in seconds) */
+    /* townengine is fixed step based, so you don't have */
+    /* TODO: actually set it */
+    float frame_duration;
 
-    /* set just once on startup */
+    /* resolution is set from config and dictates both logical and drawing space, as they're related */
+    /* even if scaling is done, game logic should never change over that */
+    Vec2i resolution;
+    Vec2i mouse_position;
+    Vec2i mouse_movement;
+
+    /* is set on startup, should be used as source of randomness */
     uint64_t random_seed;
 
-    /* this should be a multiple of the current ticks per second */
-    /* use it to simulate low framerate (e.g. at 60 tps, set to 2 for 30 fps) */
-    /* it can be changed at runtime; any resulting logic anomalies are bugs */
-    unsigned int update_multiplicity;
-
-    /* TODO: use Vec2i? */
-    int window_w;
-    int window_h;
-    int base_draw_w;
-    int base_draw_h;
-
+    /* whether debugging logic should be enabled in user code */
     bool debug;
-    bool is_running;
-    bool window_size_has_changed;
+
+    /* is set to true when state is invalidated and needs to be rebuilt */
+    /* watch for it and handle properly! */
     bool initialization_needed;
 } Context;
 
+/* when included after twn_engine_context there's an 'ctx' defined already */
 #ifndef TWN_ENGINE_CONTEXT_C_H
 TWN_API extern Context ctx;
 #endif

include/twn_draw.h

@@ -1,7 +1,7 @@
 #ifndef TWN_DRAW_H
 #define TWN_DRAW_H
 
-#include "twn_util.h"
+#include "twn_types.h"
 #include "twn_option.h"
 #include "twn_camera.h"
 #include "twn_engine_api.h"
@@ -22,6 +22,7 @@ TWN_API void draw_sprite(char const *path,
 TWN_API void draw_rectangle(Rect rect, Color color);
 
 /* pushes a filled circle onto the circle render queue */
+/* note that its edges may look jagged with a radius larger than 2048 */
 TWN_API void draw_circle(Vec2 position, float radius, Color color);
 
 /* TODO: have font optional, with something minimal coming embedded */

include/twn_engine_api.h

@@ -4,7 +4,7 @@
 #if defined(__WIN32)
 	#define TWN_API __declspec(dllexport)
 #else
-	#define TWN_API
+	#define TWN_API __attribute__((visibility("default")))
 #endif
 
 #endif

include/twn_game_api.h

@@ -1,22 +1,23 @@
 /* include this header in game code to get the usable parts of the engine */
-#ifndef GAME_API_H
-#define GAME_API_H
-
+#ifndef TWN_GAME_API_H
+#define TWN_GAME_API_H
 
+#include "twn_input.h"
 #include "twn_context.h"
 #include "twn_draw.h"
 #include "twn_audio.h"
-#include "twn_util.h"
-#include "twn_input.h"
 #include "twn_engine_api.h"
+#include "twn_util.h"
 
+#ifndef TWN_NOT_C
 
-/* sole game logic and display function.
-   all state must be used from and saved to supplied state pointer. */
-TWN_API extern void game_tick(void);
-
-/* called when application is closing. */
-TWN_API extern void game_end(void);
+   /* sole game logic and display function.
+      all state must be used from and saved to supplied state pointer. */
+   TWN_API extern void game_tick(void);
 
+   /* called when application is closing. */
+   TWN_API extern void game_end(void);
+
+#endif
 
 #endif

include/twn_input.h

@@ -1,7 +1,6 @@
 #ifndef TWN_INPUT_H
 #define TWN_INPUT_H
 
-#include "twn_config.h"
 #include "twn_types.h"
 #include "twn_engine_api.h"
 #include "twn_control.h"

include/twn_texture_modes.h

@@ -1,6 +1,9 @@
 #ifndef TWN_TEXTURES_MODES_H
 #define TWN_TEXTURES_MODES_H
 
+/* TODO: rename, as it doesn't have to be about textures only, but blending */
+/* TODO: move from public /include/ tree */
+
 /* alpha channel information */
 typedef enum TextureMode {
     TEXTURE_MODE_OPAQUE,        /* all pixels are solid */

include/twn_types.h

@@ -8,7 +8,6 @@
 
 /* a point in some space (integer) */
 typedef struct Vec2i {
-_Alignas(8)
     int32_t x;
     int32_t y;
 } Vec2i;
@@ -16,7 +15,6 @@ _Alignas(8)
 
 /* a point in some space (floating point) */
 typedef struct Vec2 {
-_Alignas(8)
     float x;
     float y;
 } Vec2;
@@ -25,7 +23,6 @@ _Alignas(8)
 /* a point in some three dimension space (floating point) */
 /* y goes up, x goes to the right */
 typedef struct Vec3 {
-_Alignas(16)
     float x;
     float y;
     float z;
@@ -35,7 +32,6 @@ _Alignas(16)
 /* a point in some three dimension space (floating point) */
 /* y goes up, x goes to the right */
 typedef struct Vec4 {
-_Alignas(16)
     float x;
     float y;
     float z;
@@ -45,7 +41,6 @@ _Alignas(16)
 
 /* 32-bit color data */
 typedef struct Color {
-_Alignas(4)
     uint8_t r;
     uint8_t g;
     uint8_t b;
@@ -55,7 +50,6 @@ _Alignas(4)
 
 /* a rectangle with the origin at the upper left (integer) */
 typedef struct Recti {
-_Alignas(16)
     int32_t x;
     int32_t y;
     int32_t w;
@@ -65,7 +59,6 @@ _Alignas(16)
 
 /* a rectangle with the origin at the upper left (floating point) */
 typedef struct Rect {
-_Alignas(16)
     float x;
     float y;
     float w;

include/twn_util.h

@@ -1,23 +1,15 @@
-#ifndef UTIL_H
-#define UTIL_H
+#ifndef TWN_UTIL_H
+#define TWN_UTIL_H
 
 #include "twn_types.h"
 #include "twn_engine_api.h"
-#include "twn_types.h"
 
 #include <stdint.h>
+#include <stddef.h>
 #include <stdbool.h>
 #include <math.h>
 
 
-/* DON'T FORGET ABOUT DOUBLE EVALUATION */
-/* YOU THINK YOU WON'T, AND THEN YOU DO */
-/* C23's typeof could fix it, so i will */
-/* leave them as macros to be replaced. */
-/* use the macro in tgmath.h for floats */
-#define MAX SDL_max
-#define MIN SDL_min
-
 #ifndef M_PI
 #define M_PI 3.14159265358979323846264338327950288   /**< pi */
 #endif
@@ -26,16 +18,13 @@
 #define DEG2RAD (M_PI / 180)
 #define RAD2DEG (180 / M_PI)
 
+#ifndef TWN_NOT_C
 
-/*                   */
-/* GENERAL UTILITIES */
-/*                   */
+    TWN_API void *cmalloc(size_t size);
+    TWN_API void *crealloc(void *ptr, size_t size);
+    TWN_API void *ccalloc(size_t num, size_t size);
 
-TWN_API void cry_impl(const char *file, const int line, const char *title, const char *text);
-#define CRY(title, text) cry_impl(__FILE__, __LINE__, title, text)
-#define CRY_SDL(title) cry_impl(__FILE__, __LINE__, title, SDL_GetError())
-#define CRY_PHYSFS(title)                                               \
-cry_impl(__FILE__, __LINE__, title, PHYSFS_getErrorByCode(PHYSFS_getLastErrorCode()))
+#endif /* TWN_NOT_C */
 
 
 TWN_API void log_info(const char *restrict format, ...);
@@ -43,18 +32,6 @@ TWN_API void log_critical(const char *restrict format, ...);
 TWN_API void log_warn(const char *restrict format, ...);
 
 
-/* for when there's absolutely no way to continue */
-TWN_API _Noreturn void die_abruptly(void);
-
-
-/* "critical" allocation functions which will log and abort() on failure. */
-/* if it is reasonable to handle this gracefully, use the standard versions. */
-/* the stb implementations will be configured to use these */
-TWN_API void *cmalloc(size_t size);
-TWN_API void *crealloc(void *ptr, size_t size);
-TWN_API void *ccalloc(size_t num, size_t size);
-
-
 /* TODO: this is why generics were invented. sorry, i'm tired today */
 TWN_API double clamp(double d, double min, double max);
 TWN_API float clampf(float f, float min, float max);

src/game_object/twn_linux_game_object.c

@@ -1,5 +1,6 @@
 #include "twn_game_object_c.h"
 #include "twn_engine_context_c.h"
+#include "twn_util_c.h"
 
 #include <x-watcher.h>
 #include <SDL2/SDL.h>
@@ -38,7 +39,7 @@ static void load_game_object(void) {
         goto ERR_OPENING_SO;
     }
 
-#pragma GCC diagnostic push 
+#pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wpedantic"
 
     game_tick_callback = (void (*)(void))dlsym(new_handle, "game_tick");
@@ -57,7 +58,7 @@ static void load_game_object(void) {
 
     handle = new_handle;
 
-    if (ctx.game.tick_count != 0)
+    if (ctx.game.frame_number != 0)
         log_info("Game object was reloaded\n");
 
     return;
@@ -82,13 +83,20 @@ static void watcher_callback(XWATCHER_FILE_EVENT event,
     (void)data;
 
     switch(event) {
+        case XWATCHER_FILE_CREATED:
         case XWATCHER_FILE_MODIFIED:
             SDL_LockMutex(lock);
-            last_tick_modified = ctx.game.tick_count;
+            last_tick_modified = ctx.game.frame_number;
             loaded_after_modification = false;
             SDL_UnlockMutex(lock);
             break;
 
+        case XWATCHER_FILE_UNSPECIFIED:
+        case XWATCHER_FILE_REMOVED:
+        case XWATCHER_FILE_OPENED:
+        case XWATCHER_FILE_ATTRIBUTES_CHANGED:
+        case XWATCHER_FILE_NONE:
+        case XWATCHER_FILE_RENAMED:
         default:
             break;
     }
@@ -129,7 +137,7 @@ bool game_object_try_reloading(void) {
 
     /* only load the modified library after some time, as compilers make a lot of modifications */
     SDL_LockMutex(lock);
-    if (ctx.game.tick_count - last_tick_modified > MODIFIED_TICKS_MERGED &&
+    if (ctx.game.frame_number - last_tick_modified > MODIFIED_TICKS_MERGED &&
             !loaded_after_modification) {
         load_game_object();
         loaded_after_modification = true;

src/game_object/twn_win32_game_object.c

@@ -1,5 +1,6 @@
 #include "twn_game_object_c.h"
 #include "twn_engine_context_c.h"
+#include "twn_util_c.h"
 
 #include <errhandlingapi.h>
 #include <libloaderapi.h>
@@ -48,7 +49,7 @@ static void load_game_object(void) {
 
     handle = new_handle;
 
-    if (ctx.game.tick_count != 0)
+    if (ctx.game.frame_number != 0)
         log_info("Game object was reloaded\n");
 
     return;

src/rendering/twn_circles.c

@@ -1,4 +1,3 @@
-#include "twn_util.h"
 #include "twn_engine_context_c.h"
 #include "twn_draw_c.h"
 #include "twn_draw.h"
@@ -22,28 +21,17 @@ void draw_circle(Vec2 position, float radius, Color color) {
     arrput(ctx.render_queue_2d, primitive);
 }
 
-/* TODO: caching and reuse scheme */
-/* vertices_out and indices_out MUST BE FREED */
+
 void create_circle_geometry(Vec2 position,
-                            Color color,
                             float radius,
                             size_t num_vertices,
-                            SDL_Vertex **vertices_out,
-                            int **indices_out)
+                            Vec2 vertices[])
 {
-    SDL_Vertex *vertices = cmalloc(sizeof *vertices * (num_vertices + 1));
-    int *indices = cmalloc(sizeof *indices * (num_vertices * 3));
-
     /* the angle (in radians) to rotate by on each iteration */
     float seg_rotation_angle = (360.0f / (float)num_vertices) * ((float)M_PI / 180);
 
-    vertices[0].position.x = (float)position.x;
-    vertices[0].position.y = (float)position.y;
-    vertices[0].color.r = color.r;
-    vertices[0].color.g = color.g;
-    vertices[0].color.b = color.b;
-    vertices[0].color.a = color.a;
-    vertices[0].tex_coord = (SDL_FPoint){ 0, 0 };
+    vertices[0].x = (float)position.x;
+    vertices[0].y = (float)position.y;
 
     /* this point will rotate around the center */
     float start_x = 0.0f - radius;
@@ -52,37 +40,13 @@ void create_circle_geometry(Vec2 position,
     for (size_t i = 1; i < num_vertices + 1; ++i) {
         float final_seg_rotation_angle = (float)i * seg_rotation_angle;
 
-        vertices[i].position.x =
-            cosf(final_seg_rotation_angle) * start_x -
-            sinf(final_seg_rotation_angle) * start_y;
-        vertices[i].position.y =
-            cosf(final_seg_rotation_angle) * start_y +
-            sinf(final_seg_rotation_angle) * start_x;
+        float c, s;
+        sincosf(final_seg_rotation_angle, &s, &c);
 
-        vertices[i].position.x += position.x;
-        vertices[i].position.y += position.y;
+        vertices[i].x = c * start_x - s * start_y;
+        vertices[i].y = c * start_y + s * start_x;
 
-        vertices[i].color.r = color.r;
-        vertices[i].color.g = color.g;
-        vertices[i].color.b = color.b;
-        vertices[i].color.a = color.a;
-
-        vertices[i].tex_coord = (SDL_FPoint){ 0, 0 };
-
-
-        size_t triangle_offset = 3 * (i - 1);
-
-        /* center point index */
-        indices[triangle_offset] = 0;
-        /* generated point index */
-        indices[triangle_offset + 1] = (int)i;
-
-        size_t index = (i + 1) % num_vertices;
-        if (index == 0)
-            index = num_vertices;
-        indices[triangle_offset + 2] = (int)index;
+        vertices[i].x += position.x;
+        vertices[i].y += position.y;
     }
-
-    *vertices_out = vertices;
-    *indices_out = indices;
 }

src/rendering/twn_draw.c

@@ -35,22 +35,6 @@ void render_queue_clear(void) {
 }
 
 
-/* rectangle */
-void draw_rectangle(Rect rect, Color color) {
-    RectPrimitive rectangle = {
-        .rect = rect,
-        .color = color,
-    };
-
-    Primitive2D primitive = {
-        .type = PRIMITIVE_2D_RECT,
-        .rect = rectangle,
-    };
-
-    arrput(ctx.render_queue_2d, primitive);
-}
-
-
 void draw_9slice(const char *texture_path, int texture_w, int texture_h, int border_thickness, Rect rect, Color color) {
     const float bt = (float)border_thickness; /* i know! */
     const float bt2 = bt * 2; /* combined size of the two borders in an axis */
@@ -197,34 +181,152 @@ static void render_2d(void) {
 
     const size_t render_queue_len = arrlenu(ctx.render_queue_2d);
 
-    size_t batch_count = 0;
+    struct Render2DInvocation {
+        Primitive2D const *primitive;
+        double layer;
+        union {
+            struct QuadBatch quad_batch;
+        };
+    };
+
+    /* first, collect all invocations, while merging into batches where applicable */
+    /* we separate into opaque and transparent ones, as it presents optimization opportunities */
+    struct Render2DInvocation *opaque_invocations = NULL;
+    struct Render2DInvocation *ghostly_invocations = NULL;
+
+    arrsetcap(opaque_invocations, render_queue_len);
+    arrsetcap(ghostly_invocations, render_queue_len);
 
     for (size_t i = 0; i < render_queue_len; ++i) {
         const Primitive2D *current = &ctx.render_queue_2d[i];
 
+        // TODO: https://gamedev.stackexchange.com/questions/101136/using-full-resolution-of-depth-buffer-for-2d-rendering
+        double const layer = ((double)((render_queue_len + 1) - i) / (double)(render_queue_len + 1)) * 0.75;
+
         switch (current->type) {
             case PRIMITIVE_2D_SPRITE: {
-                const struct SpriteBatch batch =
+                const struct QuadBatch batch =
                     collect_sprite_batch(current, render_queue_len - i);
 
-                /* TODO: what's even the point? just use OR_EQUAL comparison */
-                set_depth_range((double)batch_count / UINT16_MAX, 1.0);
-                render_sprites(current, batch);
+                struct Render2DInvocation const invocation = {
+                    .primitive = current,
+                    .quad_batch = batch,
+                    .layer = layer,
+                };
 
-                i += batch.size - 1; ++batch_count;
+                if (batch.mode == TEXTURE_MODE_GHOSTLY)
+                    arrput(ghostly_invocations, invocation);
+                else
+                    arrput(opaque_invocations, invocation);
+
+                i += batch.size - 1;
                 break;
             }
-            case PRIMITIVE_2D_RECT:
-                render_rectangle(&current->rect);
+
+            case PRIMITIVE_2D_RECT: {
+                const struct QuadBatch batch =
+                    collect_rect_batch(current, render_queue_len - i);
+
+                struct Render2DInvocation const invocation = {
+                    .primitive = current,
+                    .quad_batch = batch,
+                    .layer = layer,
+                };
+
+                if (batch.mode == TEXTURE_MODE_GHOSTLY)
+                    arrput(ghostly_invocations, invocation);
+                else
+                    arrput(opaque_invocations, invocation);
+
+                i += batch.size - 1;
                 break;
-            case PRIMITIVE_2D_CIRCLE:
-                render_circle(&current->circle);
+            }
+
+            case PRIMITIVE_2D_CIRCLE: {
+                struct Render2DInvocation const invocation = {
+                    .primitive = current,
+                    .layer = layer,
+                };
+
+                if (current->circle.color.a != 255)
+                    arrput(ghostly_invocations, invocation);
+                else
+                    arrput(opaque_invocations, invocation);
                 break;
-            case PRIMITIVE_2D_TEXT:
-                render_text(&current->text);
+            }
+
+            case PRIMITIVE_2D_TEXT: {
+                struct Render2DInvocation const invocation = {
+                    .primitive = current,
+                    .layer = layer,
+                };
+                arrput(ghostly_invocations, invocation);
                 break;
+            }
+
+            default:
+                SDL_assert(false);
         }
     }
+
+    /* first issue all opaque primitives, front-to-back */
+    for (size_t i = 0; i < arrlenu(opaque_invocations); ++i) {
+        struct Render2DInvocation const invocation = opaque_invocations[arrlenu(opaque_invocations) - 1 - i];
+
+        /* idea here is to set constant z write that moves further and further along */
+        /* with that every batch can early z reject against the previous */
+        /* additionally, it will also apply for future transparent passes, sandwitching in-between */
+        set_depth_range(invocation.layer, 1.0);
+
+        switch (invocation.primitive->type) {
+            case PRIMITIVE_2D_SPRITE: {
+                render_sprite_batch(invocation.primitive, invocation.quad_batch);
+                break;
+            }
+            case PRIMITIVE_2D_RECT: {
+                render_rect_batch(invocation.primitive, invocation.quad_batch);
+                break;
+            }
+            /* TODO: circle batching */
+            case PRIMITIVE_2D_CIRCLE:
+                render_circle(&invocation.primitive->circle);
+                break;
+            case PRIMITIVE_2D_TEXT:
+            default:
+                SDL_assert(false);
+        }
+    }
+
+    /* then issue all transparent primitives, back-to-front */
+    for (size_t i = 0; i < arrlenu(ghostly_invocations); ++i) {
+        struct Render2DInvocation const invocation = ghostly_invocations[i];
+
+        /* now we use it not for writing layers, but inferring ordering */
+        set_depth_range(invocation.layer, 1.0);
+
+        switch (invocation.primitive->type) {
+            case PRIMITIVE_2D_SPRITE: {
+                render_sprite_batch(invocation.primitive, invocation.quad_batch);
+                break;
+            }
+            case PRIMITIVE_2D_RECT: {
+                render_rect_batch(invocation.primitive, invocation.quad_batch);
+                break;
+            }
+            /* TODO: circle batching */
+            case PRIMITIVE_2D_CIRCLE:
+                render_circle(&invocation.primitive->circle);
+                break;
+            case PRIMITIVE_2D_TEXT:
+                render_text(&invocation.primitive->text);
+                break;
+            default:
+                SDL_assert(false);
+        }
+    }
+
+    arrfree(opaque_invocations);
+    arrfree(ghostly_invocations);
 }
 
 
@@ -250,33 +352,33 @@ void render(void) {
     textures_update_atlas(&ctx.texture_cache);
 
     /* fit rendering context onto the resizable screen */
-    if (ctx.game.window_size_has_changed) {
-        if ((float)ctx.game.window_w / (float)ctx.game.window_h > (float)ctx.base_render_width / (float)ctx.base_render_height) {
-            float ratio = (float)ctx.game.window_h / (float)ctx.base_render_height;
+    if (ctx.window_size_has_changed) {
+        if ((float)ctx.window_dims.x / (float)ctx.window_dims.y > (float)ctx.base_render_width / (float)ctx.base_render_height) {
+            float ratio = (float)ctx.window_dims.y / (float)ctx.base_render_height;
             int w = (int)((float)ctx.base_render_width * ratio);
             setup_viewport(
-                ctx.game.window_w / 2 - w / 2,
+                ctx.window_dims.x / 2 - w / 2,
                 0,
                 w,
-                ctx.game.window_h
+                ctx.window_dims.y
             );
         } else {
-            float ratio = (float)ctx.game.window_w / (float)ctx.base_render_width;
+            float ratio = (float)ctx.window_dims.x / (float)ctx.base_render_width;
             int h = (int)((float)ctx.base_render_height * ratio);
             setup_viewport(
                 0,
-                ctx.game.window_h / 2 - h / 2,
-                ctx.game.window_w,
+                ctx.window_dims.y / 2 - h / 2,
+                ctx.window_dims.x,
                 h
             );
         }
     }
 
-    render_space();
-    render_skybox(); /* after space, as to use depth buffer for early rejection */
-    render_2d();
-    swap_buffers();
-    clear_draw_buffer();
+    start_render_frame(); {
+        render_space();
+        render_skybox(); /* after space, as to use depth buffer for early z rejection */
+        render_2d();
+    } end_render_frame();
 }
 
 

src/rendering/twn_draw_c.h

@@ -20,7 +20,10 @@
 extern Matrix4 camera_projection_matrix;
 extern Matrix4 camera_look_at_matrix;
 
+
 #define QUAD_ELEMENT_BUFFER_LENGTH (65536 / 6)
+#define CIRCLE_VERTICES_MAX 2048
+#define CIRCLE_ELEMENT_BUFFER_LENGTH (CIRCLE_VERTICES_MAX * 3)
 
 
 typedef GLuint VertexBuffer;
@@ -126,22 +129,25 @@ void render(void);
 /* clears all render queues */
 void render_queue_clear(void);
 
+/* fills two existing arrays with the geometry data of a circle */
+/* the size of indices must be at least 3 times the number of vertices */
 void create_circle_geometry(Vec2 position,
-                            Color color,
                             float radius,
                             size_t num_vertices,
-                            SDL_Vertex **vertices_out,
-                            int **indices_out);
+                            Vec2 vertices[]);
 
-struct SpriteBatch {
+struct QuadBatch {
     size_t size;             /* how many primitives are in current batch */
-    TextureMode mode;
+    TextureMode mode;        /* how color should be applied */
     bool constant_colored;   /* whether colored batch is uniformly colored */
     bool repeat;             /* whether repeat is needed */
-} collect_sprite_batch(const Primitive2D primitives[], size_t len);
-
-void render_sprites(const Primitive2D primitives[],
-                    const struct SpriteBatch batch);
+    bool textured;
+} collect_quad_batch(const Primitive2D primitives[], size_t len);
+void render_quad_batch(const Primitive2D primitives[], struct QuadBatch batch);
+struct QuadBatch collect_sprite_batch(const Primitive2D primitives[], size_t len);
+struct QuadBatch collect_rect_batch(const Primitive2D primitives[], size_t len);
+void render_sprite_batch(const Primitive2D primitives[], struct QuadBatch batch);
+void render_rect_batch(const Primitive2D primitives[], struct QuadBatch batch);
 
 void draw_uncolored_space_traingle_batch(MeshBatch *batch,
                                          TextureKey texture_key);
@@ -160,16 +166,18 @@ void text_cache_reset_arena(TextCache *cache);
 
 VertexBuffer create_vertex_buffer(void);
 
+VertexBuffer get_scratch_vertex_array(void);
+
 void delete_vertex_buffer(VertexBuffer buffer);
 
-void specify_vertex_buffer(VertexBuffer buffer, void *data, size_t bytes);
+void specify_vertex_buffer(VertexBuffer buffer, void const *data, size_t bytes);
 
 /* uses present in 1.5 buffer mapping feature */
 VertexBufferBuilder build_vertex_buffer(VertexBuffer buffer, size_t bytes);
 
 /* collects bytes for sending to the gpu until all is pushed, which is when false is returned */
 bool push_to_vertex_buffer_builder(VertexBufferBuilder *builder,
-                                   void *bytes,
+                                   void const *bytes,
                                    size_t size);
 
 /* state */
@@ -182,7 +190,9 @@ void swap_buffers(void);
 
 void set_depth_range(double low, double high);
 
-void bind_quad_element_buffer(void);
+VertexBuffer get_quad_element_buffer(void);
+
+VertexBuffer get_circle_element_buffer(void);
 
 void render_circle(const CirclePrimitive *circle);
 
@@ -194,17 +204,17 @@ void use_2d_pipeline(void);
 
 void use_texture_mode(TextureMode mode);
 
-void finally_render_sprites(Primitive2D const primitives[],
-                            struct SpriteBatch batch,
-                            VertexBuffer buffer);
+void finally_render_quads(Primitive2D const primitives[],
+                          struct QuadBatch batch,
+                          VertexBuffer buffer);
 
-size_t get_sprite_payload_size(struct SpriteBatch batch);
+size_t get_quad_payload_size(struct QuadBatch batch);
 
-bool push_sprite_payload_to_vertex_buffer_builder(struct SpriteBatch batch,
-                                                  VertexBufferBuilder *builder,
-                                                  Vec2 v0, Vec2 v1, Vec2 v2, Vec2 v3,
-                                                  Vec2 uv0, Vec2 uv1, Vec2 uv2, Vec2 uv3,
-                                                  Color color);
+bool push_quad_payload_to_vertex_buffer_builder(struct QuadBatch batch,
+                                                VertexBufferBuilder *builder,
+                                                Vec2 v0, Vec2 v1, Vec2 v2, Vec2 v3,
+                                                Vec2 uv0, Vec2 uv1, Vec2 uv2, Vec2 uv3,
+                                                Color color);
 
 void finally_draw_uncolored_space_traingle_batch(MeshBatch const *batch,
                                                  TextureKey texture_key,
@@ -233,4 +243,8 @@ void pop_fog(void);
 
 void finally_pop_fog(void);
 
+void start_render_frame(void);
+
+void end_render_frame(void);
+
 #endif

src/rendering/twn_gl_15_gpu_texture.c

@@ -1,5 +1,5 @@
 #include "twn_gpu_texture_c.h"
-#include "twn_util.h"
+#include "twn_util_c.h"
 
 
 GPUTexture create_gpu_texture(TextureFilter filter, bool generate_mipmaps) {

src/rendering/twn_gl_15_rendering.c

@@ -1,17 +1,17 @@
 #include "twn_draw_c.h"
 #include "twn_util.h"
 #include "twn_util_c.h"
-#include "twn_config.h"
 #include "twn_engine_context_c.h"
 #include "twn_text_c.h"
+#include "twn_types.h"
 
 #include <glad/glad.h>
+#include <stb_ds.h>
 
 
+/* TODO: try using the fact we utilize edge coloring and step virtual color attributes to bogus points */
+/*       this is only doable is we take out color attribute to separate array or a portion of it */
 /* interleaved vertex array data */
-/* TODO: use int16_t for uvs */
-/* TODO: use packed types? */
-/* TODO: int16_t could be used for positioning, but we would need to have more CPU calcs */
 typedef struct ElementIndexedQuad {
     /* upper-left */
     Vec2 v0;
@@ -31,7 +31,6 @@ typedef struct ElementIndexedQuad {
     Color c3;
 } ElementIndexedQuad;
 
-
 typedef struct ElementIndexedQuadWithoutColor {
     /* upper-left */
     Vec2 v0;
@@ -48,6 +47,75 @@ typedef struct ElementIndexedQuadWithoutColor {
 } ElementIndexedQuadWithoutColor;
 
 
+typedef struct ElementIndexedQuadWithoutTexture {
+    /* upper-left */
+    Vec2 v0;
+    Color c0;
+    /* bottom-left */
+    Vec2 v1;
+    Color c1;
+    /* bottom-right */
+    Vec2 v2;
+    Color c2;
+    /* upper-right */
+    Vec2 v3;
+    Color c3;
+} ElementIndexedQuadWithoutTexture;
+
+
+typedef struct ElementIndexedQuadWithoutColorWithoutTexture {
+    /* upper-left */
+    Vec2 v0;
+    /* bottom-left */
+    Vec2 v1;
+    /* bottom-right */
+    Vec2 v2;
+    /* upper-right */
+    Vec2 v3;
+} ElementIndexedQuadWithoutColorWithoutTexture;
+
+
+typedef struct {
+    size_t offset;
+    GLenum type;
+    GLsizei stride;
+    GLuint buffer;
+    uint8_t arity; /* leave at 0 to signal pointer as unused */
+} AttributeArrayPointer;
+
+
+/* allows us to have generic way to issue draws as well as */
+/* deferring new draw calls while previous frame is still being drawn */
+typedef struct {
+    AttributeArrayPointer vertices;
+    AttributeArrayPointer texcoords;
+
+    bool constant_colored;
+    union {
+        AttributeArrayPointer colors;        
+        Color color;
+    };
+
+    bool textured, texture_repeat, uses_gpu_key;
+    TextureKey texture_key;
+    GPUTexture gpu_texture;
+
+    GLuint element_buffer;
+    GLsizei element_count;
+    GLsizei range_start, range_end;
+
+    double depth_range_low, depth_range_high;
+} DeferredCommandDraw;
+
+
+typedef struct {
+    Color color;
+    bool clear_color;
+    bool clear_depth;
+    bool clear_stencil;
+} DeferredCommandClear;
+
+
 typedef enum {
     PIPELINE_NO,
     PIPELINE_SPACE,
@@ -55,10 +123,266 @@ typedef enum {
 } Pipeline;
 
 
+typedef struct {
+    Pipeline pipeline;
+} DeferredCommandUsePipeline;
+
+
+typedef struct {
+    TextureMode mode;
+} DeferredCommandUseTextureMode;
+
+
+typedef struct {
+    double low, high;
+} DeferredCommandDepthRange;
+
+
+typedef struct {
+    enum DeferredCommandType {
+        DEFERRED_COMMAND_TYPE_DRAW,
+        DEFERRED_COMMAND_TYPE_CLEAR,
+        DEFERRED_COMMAND_TYPE_USE_PIPIELINE,
+        DEFERRED_COMMAND_TYPE_USE_TEXTURE_MODE,
+        DEFERRED_COMMAND_TYPE_DEPTH_RANGE,
+    } type;
+
+    union {
+        DeferredCommandDraw draw;
+        DeferredCommandClear clear;
+        DeferredCommandUsePipeline use_pipeline;
+        DeferredCommandUseTextureMode use_texture_mode;
+        DeferredCommandDepthRange depth_range;
+    };
+} DeferredCommand;
+
+
+static TextureMode texture_mode_last_used = -1;
 static Pipeline pipeline_last_used = PIPELINE_NO;
 
 
+/* potentially double buffered array of vertex array handles */
+/* we assume they will be refilled fully each frame */
+static size_t scratch_va_front_used, scratch_va_back_used;
+static GLuint *front_scratch_vertex_arrays, *back_scratch_vertex_arrays;
+static GLuint **current_scratch_vertex_array = &front_scratch_vertex_arrays;
+
+static void restart_scratch_vertex_arrays(void) {
+    scratch_va_front_used = 0;
+    scratch_va_back_used = 0;
+
+    if (ctx.render_double_buffered) {
+        current_scratch_vertex_array = current_scratch_vertex_array == &front_scratch_vertex_arrays ?
+                &back_scratch_vertex_arrays : &front_scratch_vertex_arrays;
+    }
+}
+
+
+GLuint get_scratch_vertex_array(void) {
+    size_t *used = current_scratch_vertex_array == &front_scratch_vertex_arrays ?
+            &scratch_va_front_used : &scratch_va_back_used;
+
+    if (arrlenu(*current_scratch_vertex_array) <= *used) {
+        GLuint handle;
+        glGenBuffers(1, &handle);
+        arrpush(*current_scratch_vertex_array, handle);
+    }
+
+    (*used)++;
+    return (*current_scratch_vertex_array)[*used - 1];
+}
+
+
+static void finally_use_2d_pipeline(void);
+static void finally_use_space_pipeline(void);
+static void finally_use_texture_mode(TextureMode mode);
+
+static DeferredCommand *deferred_commands;
+
+static void issue_deferred_draw_commands(void) {
+    for (size_t i = 0; i < arrlenu(deferred_commands); ++i) {
+        switch (deferred_commands[i].type) {
+            case DEFERRED_COMMAND_TYPE_DEPTH_RANGE: {
+                glDepthRange(deferred_commands[i].depth_range.low, deferred_commands[i].depth_range.high);
+                break;   
+            }
+            case DEFERRED_COMMAND_TYPE_CLEAR: {
+                glClearColor((1.0f / 255) * deferred_commands[i].clear.color.r,
+                             (1.0f / 255) * deferred_commands[i].clear.color.g,
+                             (1.0f / 255) * deferred_commands[i].clear.color.b,
+                             (1.0f / 255) * deferred_commands[i].clear.color.a);
+
+                /* needed as we might mess with it */
+                glDepthRange(0.0, 1.0);
+                glDepthMask(GL_TRUE);
+
+                glClear((deferred_commands[i].clear.clear_color   ? GL_COLOR_BUFFER_BIT   : 0) |
+                        (deferred_commands[i].clear.clear_depth   ? GL_DEPTH_BUFFER_BIT   : 0) |
+                        (deferred_commands[i].clear.clear_stencil ? GL_STENCIL_BUFFER_BIT : 0) );
+                break;
+            }
+
+            case DEFERRED_COMMAND_TYPE_DRAW: {
+                DeferredCommandDraw const command = deferred_commands[i].draw;
+
+                /* TODO: don't assume a single vertex array ? */
+                SDL_assert(command.vertices.arity != 0);
+                SDL_assert(command.vertices.buffer);
+                SDL_assert(command.element_count != 0);
+                SDL_assert(command.element_buffer);
+
+                glBindBuffer(GL_ARRAY_BUFFER, command.vertices.buffer);
+                glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, command.element_buffer);
+
+                glEnableClientState(GL_VERTEX_ARRAY);
+                glVertexPointer(command.vertices.arity,
+                                command.vertices.type,
+                                command.vertices.stride,
+                                (void *)command.vertices.offset);
+
+                if (command.texcoords.arity != 0) {
+                    SDL_assert(command.texcoords.buffer == command.vertices.buffer);
+
+                    glEnableClientState(GL_TEXTURE_COORD_ARRAY);
+                    glClientActiveTexture(GL_TEXTURE0);
+                    glTexCoordPointer(command.texcoords.arity,
+                                      command.texcoords.type,
+                                      command.texcoords.stride,
+                                      (void *)command.texcoords.offset);
+                }
+
+                if (command.colors.arity != 0) {
+                    SDL_assert(command.colors.buffer == command.vertices.buffer);
+
+                    glEnableClientState(GL_COLOR_ARRAY);
+                    glColorPointer(command.colors.arity,
+                                   command.colors.type,
+                                   command.colors.stride,
+                                   (void *)command.colors.offset);
+                } else if (command.constant_colored)
+                    glColor4ub(command.color.r,
+                               command.color.g,
+                               command.color.b,
+                               command.color.a);
+
+                if (command.textured) {
+                    if (command.uses_gpu_key)
+                        glBindTexture(GL_TEXTURE_2D, command.gpu_texture);
+                    else if (command.texture_repeat)
+                        textures_bind_repeating(&ctx.texture_cache, command.texture_key);
+                    else
+                        textures_bind(&ctx.texture_cache, command.texture_key);
+                }
+
+                if (command.range_start == command.range_end)
+                    glDrawElements(GL_TRIANGLES, command.element_count, GL_UNSIGNED_SHORT, NULL);
+                else
+                    glDrawRangeElements(GL_TRIANGLES,
+                                        command.range_start,
+                                        command.range_end,
+                                        command.element_count,
+                                        GL_UNSIGNED_SHORT,
+                                        NULL);
+
+                /* state clearing */
+
+                if (command.textured)
+                    glBindTexture(GL_TEXTURE_2D, 0);
+
+                if (command.colors.arity != 0)
+                    glDisableClientState(GL_COLOR_ARRAY);
+
+                if (command.texcoords.arity != 0)
+                    glDisableClientState(GL_TEXTURE_COORD_ARRAY);
+
+                glDisableClientState(GL_VERTEX_ARRAY);
+
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
+
+                break;
+            }
+
+            case DEFERRED_COMMAND_TYPE_USE_PIPIELINE: {
+                switch (deferred_commands[i].use_pipeline.pipeline) {
+                    case PIPELINE_2D:
+                        finally_use_2d_pipeline();
+                        break;
+                    case PIPELINE_SPACE:
+                        finally_use_space_pipeline();
+                        break;
+                    case PIPELINE_NO:
+                    default:
+                        SDL_assert(false);
+                }
+
+                break;
+            }
+
+            case DEFERRED_COMMAND_TYPE_USE_TEXTURE_MODE: {
+                finally_use_texture_mode(deferred_commands[i].use_texture_mode.mode);
+                break;
+            }
+
+            default:
+                SDL_assert(false);
+        }
+    }
+}
+
+
+void clear_draw_buffer(void) {
+    /* TODO: we can optimize a rectangle drawn over whole window to a clear color call*/
+
+    DeferredCommand command = {
+        .type = DEFERRED_COMMAND_TYPE_CLEAR,
+        .clear = (DeferredCommandClear) {
+            .clear_color = true,
+            .clear_depth = true,
+            .clear_stencil = true,
+            .color = (Color) { 230, 230, 230, 1 }
+        }
+    };
+
+    arrpush(deferred_commands, command);
+}
+
+
+void start_render_frame(void) {
+    clear_draw_buffer();
+}
+
+
+void end_render_frame(void) {
+    if (!ctx.render_double_buffered || ctx.game.frame_number == 0) {
+        issue_deferred_draw_commands();
+        SDL_GL_SwapWindow(ctx.window);
+        arrsetlen(deferred_commands, 0);
+        restart_scratch_vertex_arrays();
+    } else {
+        /* instead of waiting for frame to finish for the swap, we continue */
+        /* while issuing new state for the next call, but deferring any fragment emitting calls for later  */
+        /* actual swap will happen when next frame is fully finished, introducing a delay */
+        SDL_GL_SwapWindow(ctx.window);
+        issue_deferred_draw_commands();
+        restart_scratch_vertex_arrays();
+        glFlush();
+        arrsetlen(deferred_commands, 0);
+    }
+}
+
+
 void use_space_pipeline(void) {
+    DeferredCommand const command = {
+        .type = DEFERRED_COMMAND_TYPE_USE_PIPIELINE,
+        .use_pipeline = { PIPELINE_SPACE }
+    };
+
+    arrpush(deferred_commands, command);
+}
+
+
+static void finally_use_space_pipeline(void) {
     if (pipeline_last_used == PIPELINE_SPACE)
         return;
 
@@ -91,11 +415,22 @@ void use_space_pipeline(void) {
     glMatrixMode(GL_MODELVIEW);
     glLoadMatrixf(&camera_look_at_matrix.row[0].x);
 
+    texture_mode_last_used = -1;
     pipeline_last_used = PIPELINE_SPACE;
 }
 
 
 void use_2d_pipeline(void) {
+    DeferredCommand const command = {
+        .type = DEFERRED_COMMAND_TYPE_USE_PIPIELINE,
+        .use_pipeline = { PIPELINE_2D }
+    };
+
+    arrpush(deferred_commands, command);
+}
+
+
+static void finally_use_2d_pipeline(void) {
     if (pipeline_last_used == PIPELINE_SPACE) {
         glClear(GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT);
         glFlush();
@@ -127,80 +462,30 @@ void use_2d_pipeline(void) {
 
     glMatrixMode(GL_PROJECTION);
     glLoadIdentity();
-    glOrtho(0, (double)ctx.base_render_width, (double)ctx.base_render_height, 0, -1, 1);
+    glOrtho(0, (double)ctx.base_render_width, (double)ctx.base_render_height, 0, 0, 1);
 
     glMatrixMode(GL_MODELVIEW);
     glLoadIdentity();
 
+    texture_mode_last_used = -1;
     pipeline_last_used = PIPELINE_2D;
 }
 
 
-void upload_quad_vertices(Rect rect) {
-    /* client memory needs to be reachable on glDraw*, so */
-    static float vertices[6 * 2];
-
-    vertices[0]  = rect.x;          vertices[1]  = rect.y;
-    vertices[2]  = rect.x;          vertices[3]  = rect.y + rect.h;
-    vertices[4]  = rect.x + rect.w; vertices[5]  = rect.y + rect.h;
-    vertices[6]  = rect.x + rect.w; vertices[7]  = rect.y + rect.h;
-    vertices[8]  = rect.x + rect.w; vertices[9]  = rect.y;
-    vertices[10] = rect.x;          vertices[11] = rect.y;
-
-    glVertexPointer(2, GL_FLOAT, 0, (void *)&vertices);
-}
-
-
-void render_rectangle(const RectPrimitive *rectangle) {
-    glColor4ub(rectangle->color.r, rectangle->color.g,
-               rectangle->color.b, rectangle->color.a);
-
-    glEnableClientState(GL_VERTEX_ARRAY);
-    upload_quad_vertices(rectangle->rect);
-
-    glDrawArrays(GL_TRIANGLES, 0, 6);
-    glDisableClientState(GL_VERTEX_ARRAY);
-}
-
-
-void render_circle(const CirclePrimitive *circle) {
-    SDL_Vertex *vertices = NULL;
-    int *indices = NULL;
-    int num_vertices = (int)circle->radius;
-
-    create_circle_geometry(circle->position,
-                           circle->color,
-                           circle->radius,
-                           num_vertices,
-                           &vertices,
-                           &indices);
-
-    glEnableClientState(GL_VERTEX_ARRAY);
-    glVertexPointer(2,
-                    GL_FLOAT,
-                    sizeof (SDL_Vertex),
-                    &vertices->position);
-
-    glEnableClientState(GL_COLOR_ARRAY);
-    glColorPointer(4,
-                   GL_UNSIGNED_BYTE,
-                   sizeof (SDL_Vertex),
-                   &vertices->color);
-
-    glDrawElements(GL_TRIANGLES,
-                   num_vertices * 3,
-                   GL_UNSIGNED_INT,
-                   indices);
-
-    glDisableClientState(GL_COLOR_ARRAY);
-    glDisableClientState(GL_VERTEX_ARRAY);
-
-    SDL_free(vertices);
-    SDL_free(indices);
-}
-
-
 void use_texture_mode(TextureMode mode) {
+    DeferredCommand const command = {
+        .type = DEFERRED_COMMAND_TYPE_USE_TEXTURE_MODE,
+        .use_texture_mode = { mode }
+    };
+
+    arrpush(deferred_commands, command);
+}
+
+
+static void finally_use_texture_mode(TextureMode mode) {
+    if (texture_mode_last_used == mode)
+        return;
+
     static GLuint lists = 0;
     if (!lists) {
         lists = glGenLists(3);
@@ -209,7 +494,7 @@ void use_texture_mode(TextureMode mode) {
         glNewList(lists + 0, GL_COMPILE); {
             glEnable(GL_BLEND);
             glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
-            glDepthFunc(GL_ALWAYS);
+            glDepthFunc(GL_LESS);
             glDepthMask(GL_FALSE);
             glDisable(GL_ALPHA_TEST);
         } glEndList();
@@ -217,7 +502,7 @@ void use_texture_mode(TextureMode mode) {
         /* seethrough */
         glNewList(lists + 1, GL_COMPILE); {
             glDisable(GL_BLEND);
-            glDepthFunc(GL_LEQUAL);
+            glDepthFunc(GL_LESS);
             glDepthMask(GL_TRUE);
             glEnable(GL_ALPHA_TEST);
             glAlphaFunc(GL_EQUAL, 1.0f);
@@ -239,6 +524,8 @@ void use_texture_mode(TextureMode mode) {
     } else {
         glCallList(lists + 2);
     }
+
+    texture_mode_last_used = mode;
 }
 
 
@@ -258,11 +545,11 @@ VertexBufferBuilder build_vertex_buffer(VertexBuffer buffer, size_t bytes) {
 
 
 bool push_to_vertex_buffer_builder(VertexBufferBuilder *builder,
-                                   void *bytes, size_t size) {
+                                   void const *bytes, size_t size) {
     if (builder->bytes_left == 0)
         return false;
 
-    SDL_memcpy(builder->mapping, bytes, size);
+    memcpy(builder->mapping, bytes, size);
     builder->bytes_left -= size;
 
     /* trigger data send */
@@ -277,92 +564,107 @@ bool push_to_vertex_buffer_builder(VertexBufferBuilder *builder,
 }
 
 
-void finally_render_sprites(const Primitive2D primitives[],
-                            const struct SpriteBatch batch,
-                            const VertexBuffer buffer)
+void finally_render_quads(const Primitive2D primitives[],
+                          const struct QuadBatch batch,
+                          const VertexBuffer buffer)
 {
-    (void)buffer;
+    DeferredCommandDraw command = {0};
 
-    GLsizei off;
-    GLsizei voff;
-    GLsizei uvoff;
+    GLsizei off = 0, voff = 0, uvoff = 0, coff = 0;
 
-    if (!batch.constant_colored) {
+    if (!batch.constant_colored && batch.textured) {
         off   = offsetof(ElementIndexedQuad, v1);
         voff  = offsetof(ElementIndexedQuad, v0);
         uvoff = offsetof(ElementIndexedQuad, uv0);
-    } else {
+        coff  = offsetof(ElementIndexedQuad, c0);
+    } else if (batch.constant_colored && batch.textured) {
         off   = offsetof(ElementIndexedQuadWithoutColor, v1);
         voff  = offsetof(ElementIndexedQuadWithoutColor, v0);
         uvoff = offsetof(ElementIndexedQuadWithoutColor, uv0);
+    } else if (!batch.constant_colored && !batch.textured) {
+        off   = offsetof(ElementIndexedQuadWithoutTexture, v1);
+        voff  = offsetof(ElementIndexedQuadWithoutTexture, v0);
+        coff  = offsetof(ElementIndexedQuad, c0);
+    } else if (batch.constant_colored && !batch.textured) {
+        off   = offsetof(ElementIndexedQuadWithoutColorWithoutTexture, v1);
+        voff  = offsetof(ElementIndexedQuadWithoutColorWithoutTexture, v0);
     }
 
-    /* vertex specification */
-    glEnableClientState(GL_VERTEX_ARRAY);
-    glVertexPointer(2,
-                    GL_FLOAT,
-                    off,
-                    (void *)(size_t)voff);
+    command.vertices = (AttributeArrayPointer) {
+        .arity = 2,
+        .type = GL_FLOAT,
+        .stride = off,
+        .offset = voff,
+        .buffer = buffer
+    };
 
-    glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-    glClientActiveTexture(GL_TEXTURE0);
-    glTexCoordPointer(2,
-                      GL_FLOAT,
-                      off,
-                      (void *)(size_t)uvoff);
+    if (batch.textured)
+        command.texcoords = (AttributeArrayPointer) {
+            .arity = 2,
+            .type = GL_FLOAT,
+            .stride = off,
+            .offset = uvoff,
+            .buffer = buffer
+        };
 
     if (!batch.constant_colored) {
-        glEnableClientState(GL_COLOR_ARRAY);
-        glColorPointer(4,
-                       GL_UNSIGNED_BYTE,
-                       off,
-                       (void *)offsetof(ElementIndexedQuad, c0));
-    } else
-        glColor4ub(primitives[0].sprite.color.r,
-                   primitives[0].sprite.color.g,
-                   primitives[0].sprite.color.b,
-                   primitives[0].sprite.color.a);
+        command.colors = (AttributeArrayPointer) {
+            .arity = 4,
+            .type = GL_UNSIGNED_BYTE,
+            .stride = off,
+            .offset = coff,
+            .buffer = buffer
+        };
+    } else {
+        command.constant_colored = true;
+        command.color = primitives[0].sprite.color;
+    }
 
-    if (!batch.repeat)
-        textures_bind(&ctx.texture_cache, primitives->sprite.texture_key);
-    else
-        textures_bind_repeating(&ctx.texture_cache, primitives->sprite.texture_key);
+    if (batch.textured) {
+        /* TODO: bad, don't */
+        command.textured = true;
+        command.texture_key = primitives->sprite.texture_key;
+        command.texture_repeat = batch.repeat;
+    }
 
-    bind_quad_element_buffer();
+    command.element_buffer = get_quad_element_buffer();
+    command.element_count = 6 * (GLsizei)batch.size;
+    command.range_end = 6 * (GLsizei)batch.size;
 
-    glDrawElements(GL_TRIANGLES, 6 * (GLsizei)batch.size, GL_UNSIGNED_SHORT, NULL);
+    use_texture_mode(batch.mode);
 
-    /* clear the state */
-    glBufferData(GL_ARRAY_BUFFER, 0, NULL, GL_STREAM_DRAW);
-    glBindBuffer(GL_ARRAY_BUFFER, 0);
+    DeferredCommand final_command = {
+        .type = DEFERRED_COMMAND_TYPE_DRAW,
+        .draw = command
+    };
 
-    glDisableClientState(GL_TEXTURE_COORD_ARRAY);
-    glDisableClientState(GL_VERTEX_ARRAY);
-
-    if (!batch.constant_colored)
-        glDisableClientState(GL_COLOR_ARRAY);
-
-    glBindTexture(GL_TEXTURE_2D, 0);
-    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
+    arrpush(deferred_commands, final_command);
 }
 
 
-size_t get_sprite_payload_size(struct SpriteBatch batch) {
-    if (batch.constant_colored)
+size_t get_quad_payload_size(struct QuadBatch batch) {
+    if (batch.constant_colored && batch.textured)
         return sizeof (ElementIndexedQuadWithoutColor);
-    else
+    else if (!batch.constant_colored && batch.textured)
         return sizeof (ElementIndexedQuad);
+    else if (batch.constant_colored && !batch.textured)
+        return sizeof (ElementIndexedQuadWithoutTexture);
+    else if (!batch.constant_colored && !batch.textured)
+        return sizeof (ElementIndexedQuadWithoutColorWithoutTexture);
+
+    SDL_assert(false);
+    return 0;
 }
 
 
-bool push_sprite_payload_to_vertex_buffer_builder(struct SpriteBatch batch,
-                                                  VertexBufferBuilder *builder,
-                                                  Vec2 v0, Vec2 v1, Vec2 v2, Vec2 v3,
-                                                  Vec2 uv0, Vec2 uv1, Vec2 uv2, Vec2 uv3,
-                                                  Color color)
+bool push_quad_payload_to_vertex_buffer_builder(struct QuadBatch batch,
+                                                VertexBufferBuilder *builder,
+                                                Vec2 v0, Vec2 v1, Vec2 v2, Vec2 v3,
+                                                Vec2 uv0, Vec2 uv1, Vec2 uv2, Vec2 uv3,
+                                                Color color)
 {
-    if (!batch.constant_colored) {
-        ElementIndexedQuad buffer_element = {
+    if (!batch.constant_colored && batch.textured) {
+        ElementIndexedQuad const buffer_element = {
             .v0 = v0,
             .v1 = v1,
             .v2 = v2,
@@ -375,15 +677,15 @@ bool push_sprite_payload_to_vertex_buffer_builder(struct SpriteBatch batch,
 
             /* equal for all (flat shaded) */
             .c0 = color,
-            .c1 = color,
+            // .c1 = color,
             .c2 = color,
-            .c3 = color,
+            // .c3 = color,
         };
 
         return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
 
-    } else {
-        ElementIndexedQuadWithoutColor buffer_element = {
+    } else if (batch.constant_colored && batch.textured) {
+        ElementIndexedQuadWithoutColor const buffer_element = {
             .v0 = v0,
             .v1 = v1,
             .v2 = v2,
@@ -395,8 +697,37 @@ bool push_sprite_payload_to_vertex_buffer_builder(struct SpriteBatch batch,
             .uv3 = uv3,
         };
 
+        return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
+
+    } else if (!batch.constant_colored && !batch.textured) {
+        ElementIndexedQuadWithoutTexture const buffer_element = {
+            .v0 = v0,
+            .v1 = v1,
+            .v2 = v2,
+            .v3 = v3,
+
+            /* equal for all (flat shaded) */
+            .c0 = color,
+            // .c1 = color,
+            .c2 = color,
+            // .c3 = color,
+        };
+
+        return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
+
+    } else if (batch.constant_colored && !batch.textured) {
+        ElementIndexedQuadWithoutColorWithoutTexture const buffer_element = {
+            .v0 = v0,
+            .v1 = v1,
+            .v2 = v2,
+            .v3 = v3,
+        };
+
         return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
     }
+
+    SDL_assert(false);
+    return false;
 }
 
 
@@ -465,44 +796,46 @@ void finally_draw_text(FontData const *font_data,
                        Color color,
                        VertexBuffer buffer)
 {
-    (void)buffer;
+    DeferredCommandDraw command = {0};
 
-    /* vertex specification */
-    glEnableClientState(GL_VERTEX_ARRAY);
-    glVertexPointer(2,
-                    GL_FLOAT,
-                    offsetof(ElementIndexedQuadWithoutColor, v1),
-                    (void *)(size_t)offsetof(ElementIndexedQuadWithoutColor, v0));
+    command.vertices = (AttributeArrayPointer) {
+        .arity = 2,
+        .type = GL_FLOAT,
+        .stride = offsetof(ElementIndexedQuadWithoutColor, v1),
+        .offset = offsetof(ElementIndexedQuadWithoutColor, v0),
+        .buffer = buffer
+    };
 
-    glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-    glClientActiveTexture(GL_TEXTURE0);
-    glTexCoordPointer(2,
-                      GL_FLOAT,
-                      offsetof(ElementIndexedQuadWithoutColor, v1),
-                      (void *)(size_t)offsetof(ElementIndexedQuadWithoutColor, uv0));
+    command.texcoords = (AttributeArrayPointer) {
+        .arity = 2,
+        .type = GL_FLOAT,
+        .stride = offsetof(ElementIndexedQuadWithoutColor, v1),
+        .offset = offsetof(ElementIndexedQuadWithoutColor, uv0),
+        .buffer = buffer
+    };
 
-    bind_quad_element_buffer();
+    command.constant_colored = true;
+    command.color = color;
+
+    command.gpu_texture = font_data->texture;
+    command.uses_gpu_key = true;
+    command.textured = true;
+
+    command.element_buffer = get_quad_element_buffer();
+    command.element_count = 6 * (GLsizei)len;
+    command.range_end = 6 * (GLsizei)len;
 
     use_texture_mode(TEXTURE_MODE_GHOSTLY);
 
-    glBindTexture(GL_TEXTURE_2D, font_data->texture);
+    DeferredCommand final_command = {
+        .type = DEFERRED_COMMAND_TYPE_DRAW,
+        .draw = command
+    };
 
-    glColor4ub(color.r, color.g, color.b, color.a);
-
-    glDrawElements(GL_TRIANGLES, 6 * (GLsizei)len, GL_UNSIGNED_SHORT, NULL);
-
-    /* clear the state */
-    glBufferData(GL_ARRAY_BUFFER, 0, NULL, GL_STREAM_DRAW);
-    glBindBuffer(GL_ARRAY_BUFFER, 0);
-
-    glDisableClientState(GL_TEXTURE_COORD_ARRAY);
-    glDisableClientState(GL_VERTEX_ARRAY);
-
-    glBindTexture(GL_TEXTURE_2D, 0);
-    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
+    arrpush(deferred_commands, final_command);
 
     /* TODO: why doesn't it get restored if not placed here? */
-    glDepthMask(GL_TRUE);
+    // glDepthMask(GL_TRUE);
 }
 
 
@@ -531,6 +864,47 @@ static void load_cubemap_side(const char *path, GLenum target) {
     SDL_FreeSurface(surface);
 }
 
+
+void render_circle(const CirclePrimitive *circle) {
+    static Vec2 vertices[CIRCLE_VERTICES_MAX];
+    int num_vertices = CIRCLE_VERTICES_MAX - 1;
+
+    create_circle_geometry(circle->position,
+                           circle->radius,
+                           num_vertices,
+                           vertices);
+
+    VertexBuffer buffer = get_scratch_vertex_array();
+    specify_vertex_buffer(buffer, vertices, sizeof (Vec2) * num_vertices);
+
+    DeferredCommandDraw command = {0};
+
+    command.vertices = (AttributeArrayPointer) {
+        .arity = 2,
+        .type = GL_FLOAT,
+        .stride = sizeof (Vec2),
+        .offset = 0,
+        .buffer = buffer
+    };
+
+    command.constant_colored = true;
+    command.color = circle->color;
+
+    command.element_buffer = get_circle_element_buffer();
+    command.element_count = num_vertices * 3;
+    command.range_end = num_vertices * 3;
+
+    use_texture_mode(circle->color.a == 255 ? TEXTURE_MODE_OPAQUE : TEXTURE_MODE_GHOSTLY);
+
+    DeferredCommand final_command = {
+        .type = DEFERRED_COMMAND_TYPE_DRAW,
+        .draw = command
+    };
+
+    arrpush(deferred_commands, final_command);
+}
+
+
 void finally_render_skybox(char *paths) {
     static GLuint cubemap = 0;
     static char *paths_cache = NULL;
@@ -697,3 +1071,16 @@ void finally_apply_fog(float start, float end, float density, Color color) {
 void finally_pop_fog(void) {
     glDisable(GL_FOG);
 }
+
+
+void set_depth_range(double low, double high) {
+    DeferredCommand const command = {
+        .type = DEFERRED_COMMAND_TYPE_DEPTH_RANGE,
+        .depth_range = {
+            .low = low,
+            .high = high
+        }
+    };
+
+    arrpush(deferred_commands, command);
+}

src/rendering/twn_gl_any_rendering.c

@@ -1,6 +1,6 @@
 #include "twn_draw_c.h"
 #include "twn_engine_context_c.h"
-#include "twn_util.h"
+#include "twn_util_c.h"
 
 #ifdef EMSCRIPTEN
 #include <GLES2/gl2.h>
@@ -26,65 +26,66 @@ void delete_vertex_buffer(VertexBuffer buffer) {
 }
 
 
-void specify_vertex_buffer(VertexBuffer buffer, void *data, size_t bytes) {
+void specify_vertex_buffer(VertexBuffer buffer, void const *data, size_t bytes) {
     glBindBuffer(GL_ARRAY_BUFFER, buffer);
     glBufferData(GL_ARRAY_BUFFER, bytes, data, GL_STREAM_DRAW);
 }
 
 
-void bind_quad_element_buffer(void) {
-    static GLuint buffer = 0;
+VertexBuffer get_quad_element_buffer(void) {
+    static VertexBuffer buffer = 0;
 
     /* it's only generated once at runtime */
+    /* TODO: use builder interface, not direct calls (glMapBuffer isn't portable) */
     if (buffer == 0) {
-        glGenBuffers(1, &buffer);
-        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer);
-        glBufferData(GL_ELEMENT_ARRAY_BUFFER,
-                     QUAD_ELEMENT_BUFFER_LENGTH * 6 * sizeof(uint16_t),
-                     NULL,
-                     GL_STATIC_DRAW);
+        buffer = create_vertex_buffer();
+        VertexBufferBuilder builder = build_vertex_buffer(buffer, sizeof (GLshort) * QUAD_ELEMENT_BUFFER_LENGTH * 6 );
 
-        uint16_t *const indices = glMapBuffer(GL_ELEMENT_ARRAY_BUFFER,
-                                              GL_WRITE_ONLY);
-        if (!indices)
-            CRY("Quad indices generation", "glMapBuffer() failed");
+        for (size_t i = 0; i < QUAD_ELEMENT_BUFFER_LENGTH; ++i) {
+            GLshort indices[6];
+            indices[0] = (GLshort)(i * 4 + 0);
+            indices[1] = (GLshort)(i * 4 + 1);
+            indices[2] = (GLshort)(i * 4 + 2);
+            indices[3] = (GLshort)(i * 4 + 2);
+            indices[4] = (GLshort)(i * 4 + 3);
+            indices[5] = (GLshort)(i * 4 + 0);
 
-        for (uint16_t i = 0; i < QUAD_ELEMENT_BUFFER_LENGTH; ++i) {
-            indices[i * 6 + 0] = (uint16_t)(i * 4 + 0);
-            indices[i * 6 + 1] = (uint16_t)(i * 4 + 1);
-            indices[i * 6 + 2] = (uint16_t)(i * 4 + 2);
-            indices[i * 6 + 3] = (uint16_t)(i * 4 + 2);
-            indices[i * 6 + 4] = (uint16_t)(i * 4 + 3);
-            indices[i * 6 + 5] = (uint16_t)(i * 4 + 0);
+            push_to_vertex_buffer_builder(&builder, indices, sizeof indices);
         }
+    }
 
-        glUnmapBuffer(GL_ELEMENT_ARRAY_BUFFER);
+    SDL_assert_always(buffer);
 
-    } else
-        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer);
+    return buffer;
 }
 
 
-void clear_draw_buffer(void) {
-    /* TODO: we can optimize a rectangle drawn over whole window to a clear color call*/
-    glClearColor((1.0f / 255) * 230,
-                 (1.0f / 255) * 230,
-                 (1.0f / 255) * 230, 1);
+VertexBuffer get_circle_element_buffer(void) {
+    static VertexBuffer buffer = 0;
 
-    /* TODO: don't clear color when skybox is applied? */
-    /*       for that window should match framebuffer */
-    /*       also it is driver dependent, from what i can gather */
-    glClear(GL_COLOR_BUFFER_BIT |
-            GL_DEPTH_BUFFER_BIT |
-            GL_STENCIL_BUFFER_BIT);
-}
-
-
-void swap_buffers(void) {
-    SDL_GL_SwapWindow(ctx.window);
-}
-
-
-void set_depth_range(double low, double high) {
-    glDepthRange(low, high);
+    if (buffer == 0) {
+        buffer = create_vertex_buffer();
+        VertexBufferBuilder builder = build_vertex_buffer(buffer, sizeof (GLshort) * CIRCLE_ELEMENT_BUFFER_LENGTH);
+
+        for (size_t i = 1; i < CIRCLE_VERTICES_MAX; ++i) {
+            /* first one is center point index, always zero */
+            GLshort indices[3];
+
+            indices[0] = 0;
+
+            /* generated point index */
+            indices[1] = (GLshort)i;
+
+            size_t index = (i + 1) % (CIRCLE_VERTICES_MAX - 1);
+            if (index == 0) /* don't use center for outer ring */
+                index = (CIRCLE_VERTICES_MAX - 1);
+            indices[2] = (GLshort)index;
+
+            push_to_vertex_buffer_builder(&builder, indices, sizeof indices);
+        }
+    }
+
+    SDL_assert_always(buffer);
+
+    return buffer;
 }

src/rendering/twn_quads.c

@@ -0,0 +1,30 @@
+#include "twn_draw_c.h"
+
+#include <stddef.h>
+
+
+struct QuadBatch collect_quad_batch(const Primitive2D primitives[], size_t len) {
+    if (primitives[0].type == PRIMITIVE_2D_SPRITE)
+        return collect_sprite_batch(primitives, len);
+    else if (primitives[0].type == PRIMITIVE_2D_RECT)
+        return collect_rect_batch(primitives, len);
+    else
+        SDL_assert(false);
+
+    return (struct QuadBatch){0};
+}
+
+
+/* assumes that orthogonal matrix setup is done already */
+void render_quad_batch(const Primitive2D primitives[],
+                       const struct QuadBatch batch)
+{
+    if (primitives[0].type == PRIMITIVE_2D_SPRITE)
+        render_sprite_batch(primitives, batch);
+    else if (primitives[0].type == PRIMITIVE_2D_RECT)
+        render_rect_batch(primitives, batch);
+    else
+        SDL_assert(false);
+
+    return (struct QuadBatch){0};
+}

src/rendering/twn_rects.c

@@ -0,0 +1,102 @@
+#include "twn_draw.h"
+#include "twn_draw_c.h"
+#include "twn_engine_context_c.h"
+#include "twn_util.h"
+#include "twn_util_c.h"
+#include "twn_textures_c.h"
+#include "twn_option.h"
+
+#include <stb_ds.h>
+
+#include <stdbool.h>
+#include <stddef.h>
+
+
+void draw_rectangle(Rect rect, Color color) {
+    RectPrimitive rectangle = {
+        .rect = rect,
+        .color = color,
+    };
+
+    Primitive2D primitive = {
+        .type = PRIMITIVE_2D_RECT,
+        .rect = rectangle,
+    };
+
+    arrput(ctx.render_queue_2d, primitive);
+}
+
+
+struct QuadBatch collect_rect_batch(const Primitive2D primitives[], size_t len) {
+    SDL_assert(primitives[0].type == PRIMITIVE_2D_RECT);
+    SDL_assert(primitives && len != 0);
+
+    struct QuadBatch batch = {
+        .mode = primitives[0].rect.color.a == 255 ? TEXTURE_MODE_OPAQUE : TEXTURE_MODE_GHOSTLY,
+        .constant_colored = true,
+    };
+
+    const uint32_t uniform_color = *(const uint32_t *)&primitives[0].rect.color;
+
+    /* batch size is clamped so that reallocated short indices could be used */
+    if (len >= QUAD_ELEMENT_BUFFER_LENGTH)
+        len = QUAD_ELEMENT_BUFFER_LENGTH;
+
+    for (size_t i = 0; i < len; ++i) {
+        const Primitive2D *const current = &primitives[i];
+
+        /* don't touch things other than rectangles */
+        if (current->type != PRIMITIVE_2D_RECT)
+            break;
+
+        /* only collect the same blend modes */
+        if ((current->rect.color.a == 255 ? TEXTURE_MODE_OPAQUE : TEXTURE_MODE_GHOSTLY) != batch.mode)
+            break;
+
+        /* if all are modulated the same we can skip sending the color data */
+        if (*(const uint32_t *)&current->rect.color != uniform_color)
+            batch.constant_colored = false;
+
+        ++batch.size;
+    }
+
+    return batch;
+}
+
+
+/* assumes that orthogonal matrix setup is done already */
+void render_rect_batch(const Primitive2D primitives[],
+                       const struct QuadBatch batch)
+{
+    SDL_assert(primitives && batch.size != 0);
+    SDL_assert(primitives[0].type == PRIMITIVE_2D_RECT);
+
+    /* single vertex array is used for every batch with NULL glBufferData() trick at the end */
+    VertexBuffer const vertex_array = get_scratch_vertex_array();
+
+    use_texture_mode(batch.mode);
+
+    /* vertex population over a vertex buffer builder interface */
+    {
+        VertexBufferBuilder payload = build_vertex_buffer(vertex_array, get_quad_payload_size(batch) * batch.size);
+
+        for (size_t i = 0; i < batch.size; ++i) {
+            /* render opaques front to back, to gain benefit of an early z rejection */
+            const size_t cur = batch.mode == TEXTURE_MODE_GHOSTLY ? i : batch.size - i - 1;
+            const RectPrimitive rect = primitives[cur].rect;
+
+            Vec2 v0 = { rect.rect.x,          rect.rect.y };
+            Vec2 v1 = { rect.rect.x,          rect.rect.y + rect.rect.h };
+            Vec2 v2 = { rect.rect.x + rect.rect.w, rect.rect.y + rect.rect.h };
+            Vec2 v3 = { rect.rect.x + rect.rect.w, rect.rect.y };
+
+            push_quad_payload_to_vertex_buffer_builder(
+                batch, &payload,
+                v0, v1, v2, v3,
+                (Vec2){0}, (Vec2){0}, (Vec2){0}, (Vec2){0},
+                rect.color);
+        }
+    }
+
+    finally_render_quads(primitives, batch, vertex_array);
+}

src/rendering/twn_skybox.c

@@ -3,7 +3,7 @@
 
 #include <SDL2/SDL.h>
 
-char *paths_in_use;
+static char *paths_in_use;
 
 void draw_skybox(const char *paths) {
     if (paths_in_use && SDL_strcmp(paths, paths_in_use) == 0)

src/rendering/twn_sprites.c

@@ -60,15 +60,18 @@ void draw_sprite_args(const DrawSpriteArgs args) {
 }
 
 
-struct SpriteBatch collect_sprite_batch(const Primitive2D primitives[], size_t len) {
-    /* assumes that first primitive is already a sprite */
+struct QuadBatch collect_sprite_batch(const Primitive2D primitives[], size_t len) {
+    SDL_assert(primitives[0].type == PRIMITIVE_2D_SPRITE);
+    SDL_assert(primitives && len != 0);
+
     const uint16_t texture_key_id = primitives[0].sprite.texture_key.id;
     const int atlas_id = textures_get_atlas_id(&ctx.texture_cache, primitives[0].sprite.texture_key);
 
-    struct SpriteBatch batch = {
+    struct QuadBatch batch = {
         .mode = textures_get_mode(&ctx.texture_cache, primitives[0].sprite.texture_key),
         .constant_colored = true,
         .repeat = primitives[0].sprite.repeat,
+        .textured = true,
     };
 
     const uint32_t uniform_color = *(const uint32_t *)&primitives[0].sprite.color;
@@ -116,35 +119,40 @@ struct SpriteBatch collect_sprite_batch(const Primitive2D primitives[], size_t l
 
 
 /* assumes that orthogonal matrix setup is done already */
-void render_sprites(const Primitive2D primitives[],
-                    const struct SpriteBatch batch)
+void render_sprite_batch(const Primitive2D primitives[],
+                         const struct QuadBatch batch)
 {
-    /* single vertex array is used for every batch with NULL glBufferData() trick at the end */
-    static VertexBuffer vertex_array = 0;
-    if (vertex_array == 0)
-        vertex_array = create_vertex_buffer();
+    SDL_assert(primitives && batch.size != 0);
+    SDL_assert(primitives[0].type == PRIMITIVE_2D_SPRITE);
 
-    use_texture_mode(batch.mode);
+    VertexBuffer const vertex_array = get_scratch_vertex_array();
 
     const Rect dims =
         textures_get_dims(&ctx.texture_cache, primitives->sprite.texture_key);
 
+    /* cached srcrect */
+    Rect cached_srcrect;
+    TextureKey cached_srcrect_key = TEXTURE_KEY_INVALID;
+
     /* vertex population over a vertex buffer builder interface */
     {
-        VertexBufferBuilder payload = build_vertex_buffer(vertex_array, get_sprite_payload_size(batch) * batch.size);
+        VertexBufferBuilder payload = build_vertex_buffer(vertex_array, get_quad_payload_size(batch) * batch.size);
 
         for (size_t i = 0; i < batch.size; ++i) {
-            /* render opaques front to back */
+            /* render opaques front to back, to gain benefit of an early z rejection */
             const size_t cur = batch.mode == TEXTURE_MODE_GHOSTLY ? i : batch.size - i - 1;
             const SpritePrimitive sprite = primitives[cur].sprite;
 
-            /* TODO: try caching it */
-            const Rect srcrect =
-                textures_get_srcrect(&ctx.texture_cache, primitives[cur].sprite.texture_key);
+            if (primitives[cur].sprite.texture_key.id != cached_srcrect_key.id) {
+                cached_srcrect = textures_get_srcrect(&ctx.texture_cache, primitives[cur].sprite.texture_key);
+                cached_srcrect_key = primitives[cur].sprite.texture_key;
+            }
+
+            Rect const srcrect = cached_srcrect;
 
             Vec2 uv0, uv1, uv2, uv3;
 
-            if (!sprite.repeat) {
+            if (!batch.repeat) {
                 const float wr = srcrect.w / dims.w;
                 const float hr = srcrect.h / dims.h;
                 const float xr = srcrect.x / dims.w;
@@ -199,9 +207,14 @@ void render_sprites(const Primitive2D primitives[],
                 v2 = (Vec2){ sprite.rect.x + sprite.rect.w, sprite.rect.y + sprite.rect.h };
                 v3 = (Vec2){ sprite.rect.x + sprite.rect.w, sprite.rect.y };
 
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wfloat-equal"
+
             } else if (sprite.rect.w == sprite.rect.h) {
                 /* rotated square case */
 
+#pragma GCC diagnostic pop
+
                 const Vec2 c = frect_center(sprite.rect);
                 const Vec2 t = fast_cossine(sprite.rotation + (float)M_PI_4);
                 const Vec2 d = {
@@ -228,9 +241,9 @@ void render_sprites(const Primitive2D primitives[],
                 v3 = (Vec2){ c.x + t.x * +h.x - t.y * -h.y, c.y + t.y * +h.x + t.x * -h.y };
             }
 
-            push_sprite_payload_to_vertex_buffer_builder(batch, &payload, v0, v1, v2, v3, uv0, uv1, uv2, uv3, sprite.color);
+            push_quad_payload_to_vertex_buffer_builder(batch, &payload, v0, v1, v2, v3, uv0, uv1, uv2, uv3, sprite.color);
         }
     }
 
-    finally_render_sprites(primitives, batch, vertex_array);
+    finally_render_quads(primitives, batch, vertex_array);
 }

src/rendering/twn_text.c

@@ -1,11 +1,11 @@
 #include "twn_draw_c.h"
 #include "twn_draw.h"
 #include "twn_util.h"
-#include "twn_config.h"
+#include "twn_util_c.h"
 #include "twn_engine_context_c.h"
 
 #include <stb_truetype.h>
-
+#include <stb_ds.h>
 
 #define ASCII_START 32
 #define ASCII_END 128
@@ -31,6 +31,19 @@ typedef struct StringArena {
 static StringArena string_arena;
 
 
+typedef struct FontFileBuffer {
+    size_t len;
+    unsigned char *buffer;
+} FontFileBuffer;
+
+typedef struct FontFileCacheItem {
+    char *key;
+    FontFileBuffer value;
+} FontFileCacheItem;
+
+static FontFileCacheItem *font_file_cache_hash;
+
+
 static void string_arena_init(StringArena *arena) {
     arena->head = cmalloc(sizeof *arena->head);
     arena->current_block = arena->head;
@@ -102,13 +115,23 @@ static FontData *text_load_font_data(const char *path, int height_px) {
     unsigned char* bitmap = ccalloc(ctx.font_texture_size * ctx.font_texture_size, 1);
 
     {
-        unsigned char *buf = NULL; 
-        int64_t buf_len = file_to_bytes(path, &buf);
+        unsigned char *buf = NULL;
+        int64_t buf_len = 0;
+
+        /* if the file was already loaded just get it */
+        FontFileCacheItem *font_file_ptr = shgetp_null(font_file_cache_hash, path);
+        if (font_file_ptr != NULL) {
+            buf = font_file_ptr->value.buffer;
+            buf_len = font_file_ptr->value.len;
+        } else {
+            buf_len = file_to_bytes(path, &buf);
+            FontFileBuffer buffer = { buf_len, buf };
+            shput(font_file_cache_hash, path, buffer);
+        }
+
         stbtt_InitFont(&font_data->info, buf, stbtt_GetFontOffsetForIndex(buf, 0));
         
         /* might as well get these now, for later */
-        font_data->file_bytes = buf;
-        font_data->file_bytes_len = buf_len;
         font_data->scale_factor = stbtt_ScaleForPixelHeight(&font_data->info, (float)height_px);
         stbtt_GetFontVMetrics(
             &font_data->info,
@@ -142,16 +165,13 @@ static FontData *text_load_font_data(const char *path, int height_px) {
 
 
 static void text_destroy_font_data(FontData *font_data) {
-    SDL_free(font_data->file_bytes);
     delete_gpu_texture(font_data->texture);
     SDL_free(font_data);
 }
 
 
 static void text_draw_with(FontData* font_data, char* text, Vec2 position, Color color) {
-    VertexBuffer vertex_array = 0;
-    if (vertex_array == 0)
-        vertex_array = create_vertex_buffer();
+    VertexBuffer const vertex_array = get_scratch_vertex_array();
 
     const size_t len = SDL_strlen(text);
 
@@ -192,7 +212,7 @@ static void text_draw_with(FontData* font_data, char* text, Vec2 position, Color
 
 
 static void ensure_font_cache(const char *font_path, int height_px) {
-    /* HACK: stupid, bad, don't do this */
+    /* HACK: don't */
     bool is_cached = false;
     for (size_t i = 0; i < arrlenu(ctx.text_cache.data); ++i) {
         FontData *font_data = ctx.text_cache.data[i];
@@ -229,6 +249,8 @@ void render_text(const TextPrimitive *text) {
 void text_cache_init(TextCache *cache) {
     arrsetlen(cache->data, 0);
     string_arena_init(&string_arena);
+
+    sh_new_arena(font_file_cache_hash);
 }
 
 
@@ -237,8 +259,13 @@ void text_cache_deinit(TextCache *cache) {
         text_destroy_font_data(ctx.text_cache.data[i]);
     }
 
-    arrfree(cache->data);
+    for (size_t i = 0; i < shlenu(font_file_cache_hash); ++i) {
+        SDL_free(font_file_cache_hash[i].value.buffer);
+    }
+    shfree(font_file_cache_hash);
+
     string_arena_deinit(&string_arena);
+    arrfree(cache->data);
 }
 
 

src/rendering/twn_text_c.h

@@ -12,14 +12,16 @@
 #define ASCII_END 128
 #define NUM_DISPLAY_ASCII ((ASCII_END - ASCII_START) + 1)
 
+#define TEXT_FONT_TEXTURE_SIZE_DEFAULT 2048
+#define TEXT_FONT_OVERSAMPLING_DEFAULT 4
+#define TEXT_FONT_FILTERING_DEFAULT TEXTURE_FILTER_LINEAR
+
 
 typedef struct FontData {
     stbtt_packedchar char_data[NUM_DISPLAY_ASCII];
     stbtt_fontinfo info;
 
     const char *file_path;
-    unsigned char *file_bytes;
-    size_t file_bytes_len;
 
     GPUTexture texture;
 

src/rendering/twn_triangles.c

@@ -35,7 +35,7 @@ void draw_triangle(const char *path,
         .uv2 = uv2,
     }};
 
-    union UncoloredSpaceTriangle *triangles = (union UncoloredSpaceTriangle *)batch_p->value.primitives;
+    union UncoloredSpaceTriangle *triangles = (union UncoloredSpaceTriangle *)(void *)batch_p->value.primitives;
 
     arrpush(triangles, triangle);
     batch_p->value.primitives = (uint8_t *)triangles;
@@ -45,9 +45,7 @@ void draw_triangle(const char *path,
 void draw_uncolored_space_traingle_batch(struct MeshBatch *batch,
                                          TextureKey texture_key)
 {
-    static VertexBuffer vertex_array = 0;
-    if (vertex_array == 0)
-        vertex_array = create_vertex_buffer();
+    VertexBuffer const vertex_array = get_scratch_vertex_array();
 
     const size_t primitives_len = arrlenu(batch->primitives);
 
@@ -66,7 +64,7 @@ void draw_uncolored_space_traingle_batch(struct MeshBatch *batch,
     /* update pixel-based uvs to correspond with texture atlases */
     for (size_t i = 0; i < primitives_len; ++i) {
         struct UncoloredSpaceTrianglePayload *payload =
-            &((union UncoloredSpaceTriangle *)batch->primitives)[i].payload;
+            &((union UncoloredSpaceTriangle *)(void *)batch->primitives)[i].payload;
 
         payload->uv0.x = xr + ((float)payload->uv0.x / srcrect.w) * wr;
         payload->uv0.y = yr + ((float)payload->uv0.y / srcrect.h) * hr;

src/system/linux/twn_elf.c

@@ -7,7 +7,6 @@
 #include <sys/auxv.h>
 #include <elf.h>
 #include <linux/limits.h>
-#define __USE_GNU
 #include <dlfcn.h>
 
 #include <SDL2/SDL.h>

src/twn_amalgam.c

@@ -0,0 +1,20 @@
+/* a technique for faster compilation */
+/* it includes all non-optional .c files directly in a single compilation unit */
+
+#include "twn_audio.c"
+#include "twn_camera.c"
+#include "twn_context.c"
+#include "twn_input.c"
+#include "twn_loop.c"
+#include "twn_main.c"
+#include "twn_textures.c"
+#include "twn_util.c"
+
+#include "rendering/twn_circles.c"
+#include "rendering/twn_draw.c"
+#include "rendering/twn_fog.c"
+#include "rendering/twn_skybox.c"
+#include "rendering/twn_sprites.c"
+#include "rendering/twn_rects.c"
+#include "rendering/twn_text.c"
+#include "rendering/twn_triangles.c"

src/twn_audio.c

@@ -1,7 +1,7 @@
 #include "twn_audio_c.h"
-#include "twn_config.h"
 #include "twn_engine_context_c.h"
 #include "twn_util.h"
+#include "twn_util_c.h"
 
 #include <SDL2/SDL.h>
 #include <stb_ds.h>
@@ -126,6 +126,8 @@ static union AudioContext init_audio_context(const char *path, AudioFileType typ
         };
     }
 
+    case AUDIO_FILE_TYPE_UNKNOWN:
+    case AUDIO_FILE_TYPE_COUNT:
     default:
         CRY("Audio error", "Unhandled audio format (in init)");
         return (union AudioContext){0};
@@ -147,6 +149,8 @@ static void repeat_audio(AudioChannel *channel) {
         break;
     }
 
+    case AUDIO_FILE_TYPE_UNKNOWN:
+    case AUDIO_FILE_TYPE_COUNT:
     default:
         CRY("Audio error", "Unhandled audio format (in repeat)");
         break;
@@ -219,6 +223,8 @@ TWN_API void audio_set(const char *channel, AudioParam param, float value) {
         }
         pair->value.panning = value;
         break;
+    default:
+        CRY("Audio channel parameter setting failed", "Invalid parameter enum given");
     }
 }
 
@@ -229,8 +235,8 @@ static void audio_mixin_streams(const AudioChannel *channel,
                                 uint8_t *restrict b,
                                 size_t frames)
 {
-    float *const sa = (float *)a;
-    float *const sb = (float *)b;
+    float *const sa = (float *)(void *)a;
+    float *const sb = (float *)(void *)b;
 
     const float left_panning = fminf(fabsf(channel->panning - 1.0f), 1.0f);
     const float right_panning = fminf(fabsf(channel->panning + 1.0f), 1.0f);
@@ -321,6 +327,8 @@ static void audio_sample_and_mixin_channel(const AudioChannel *channel,
         break;
     }
 
+    case AUDIO_FILE_TYPE_UNKNOWN:
+    case AUDIO_FILE_TYPE_COUNT:
     default:
         CRY("Audio error", "Unhandled audio format (in sampling)");
         break;

src/twn_audio_c.h

@@ -13,6 +13,9 @@
 #include <stdint.h>
 
 
+#define AUDIO_FREQUENCY 48000
+
+
 typedef enum AudioFileType {
     AUDIO_FILE_TYPE_OGG,
     AUDIO_FILE_TYPE_XM,

src/twn_camera.c

@@ -1,5 +1,4 @@
 #include "twn_camera.h"
-#include "twn_config.h"
 #include "twn_vec.h"
 #include "twn_engine_context_c.h"
 

src/twn_engine_context_c.h

@@ -17,6 +17,8 @@
 
 typedef struct EngineContext {
     /* user code facing context */
+    Context game_copy;
+    /* engine-side context, copied to `game_copy` before every frame */
     Context game;
 
     InputState input;
@@ -27,6 +29,8 @@ typedef struct EngineContext {
     /* where the app was run from, used as the root for packs */
     char *base_dir;
 
+    Vec2i window_dims;
+
     /* configuration */
     toml_table_t *config_table;
     int64_t base_render_width;
@@ -52,6 +56,7 @@ typedef struct EngineContext {
     uint8_t audio_stream_channel_count;
 
     /* main loop machinery */
+    int64_t delta_time; /* preserves real time frame delta with no manipilation */
     int64_t clocks_per_second;
     int64_t prev_frame_time;
     int64_t desired_frametime;  /* how long one tick should be */
@@ -59,12 +64,20 @@ typedef struct EngineContext {
     int64_t delta_averager_residual;
     int64_t time_averager[4];
 
+    /* this should be a multiple of the current ticks per second */
+    /* use it to simulate low framerate (e.g. at 60 tps, set to 2 for 30 fps) */
+    /* it can be changed at runtime; any resulting logic anomalies are bugs */
+    uint32_t update_multiplicity;
+
     SDL_GLContext *gl_context;
     SDL_Window *window;
     uint32_t window_id;
 
+    bool is_running;
+    bool window_size_has_changed;
     bool resync_flag;
     bool was_successful;
+    bool render_double_buffered;
 } EngineContext;
 
 /* TODO: does it need to be marked with TWN_API? */

src/twn_input.c

@@ -15,6 +15,12 @@ static void update_action_pressed_state(InputState *input, Action *action) {
         switch (action->bindings[i].source) {
         case BUTTON_SOURCE_NOT_SET:
             break;
+        case BUTTON_SOURCE_KEYBOARD_CHARACTER:
+            CRY("Action pressed state updated failed", "BUTTON_SOURCE_KEYBOARD_CHARACTER isn't handled");
+            break;
+        case BUTTON_SOURCE_GAMEPAD:
+            CRY("Action pressed state updated failed", "BUTTON_SOURCE_GAMEPAD isn't handled");
+            break;
         case BUTTON_SOURCE_KEYBOARD_PHYSICAL:
             /* not pressed */
             if (input->keyboard_state[action->bindings[i].code.scancode] == 0) {
@@ -98,6 +104,8 @@ static void input_bind_code_to_action(InputState *input,
         case BUTTON_SOURCE_MOUSE:
             is_already_bound = binding->code.mouse_button == code.mouse_button;
             break;
+        default:
+            SDL_assert(false);
         }
 
         if (is_already_bound) {
@@ -156,6 +164,8 @@ static void input_unbind_code_from_action(InputState *input,
         case BUTTON_SOURCE_MOUSE:
             is_bound = binding->code.mouse_button == code.mouse_button;
             break;
+        default:
+            SDL_assert(false);
         }
 
         /* stop early, this won't change */
@@ -194,8 +204,8 @@ void input_state_update(InputState *input) {
     SDL_GetRelativeMouseState(&input->mouse_relative_position.x,
                               &input->mouse_relative_position.y);
 
-    ctx.game.mouse_window_position = input->mouse_window_position;
-    ctx.game.mouse_relative_position = input->mouse_relative_position;
+    ctx.game.mouse_position = input->mouse_window_position;
+    ctx.game.mouse_movement = input->mouse_relative_position;
 
     for (size_t i = 0; i < shlenu(input->action_hash); ++i) {
         Action *action = &input->action_hash[i].value;
@@ -220,7 +230,7 @@ void input_bind_action_control(char const *action_name,
         input_bind_code_to_action(&ctx.input,
                                   action_name,
                                   BUTTON_SOURCE_MOUSE,
-                                  (union ButtonCode) { .mouse_button = SDL_BUTTON(mouse_button)});
+                                  (union ButtonCode) { .mouse_button = (uint8_t)SDL_BUTTON(mouse_button)});
     } else
         log_warn("(%s) Invalid control value given: %i.", __func__, control);
 }
@@ -242,7 +252,7 @@ void input_unbind_action_control(char const *action_name,
         input_unbind_code_from_action(&ctx.input,
                                       action_name,
                                       BUTTON_SOURCE_MOUSE,
-                                      (union ButtonCode) { .mouse_button = SDL_BUTTON(mouse_button)});
+                                      (union ButtonCode) { .mouse_button = (uint8_t)SDL_BUTTON(mouse_button)});
     } else
         log_warn("(%s) Invalid control value given: %i.", __func__, control);
 }

src/twn_input_c.h

@@ -7,6 +7,9 @@
 #include <SDL2/SDL.h>
 
 
+#define KEYBIND_SLOTS_DEFAULT 3
+
+
 union ButtonCode {
     SDL_Scancode scancode;
     SDL_Keycode keycode;

src/twn_loop.c

@@ -2,6 +2,7 @@
 #include "twn_engine_context_c.h"
 #include "twn_input_c.h"
 #include "twn_util.h"
+#include "twn_util_c.h"
 #include "twn_game_object_c.h"
 #include "twn_audio_c.h"
 #include "twn_textures_c.h"
@@ -22,6 +23,10 @@
 #include <limits.h>
 
 
+#define TICKS_PER_SECOND_DEFAULT 60
+#define PACKAGE_EXTENSION "btw"
+
+
 static int event_callback(void *userdata, SDL_Event *event) {
     (void)userdata;
 
@@ -32,8 +37,8 @@ static int event_callback(void *userdata, SDL_Event *event) {
 
         switch (event->window.event) {
         case SDL_WINDOWEVENT_SIZE_CHANGED:
-            ctx.game.window_w = event->window.data1;
-            ctx.game.window_h = event->window.data2;
+            ctx.window_dims.x = event->window.data1;
+            ctx.window_dims.y = event->window.data2;
             ctx.resync_flag = true;
             break;
 
@@ -55,12 +60,12 @@ static int event_callback(void *userdata, SDL_Event *event) {
 static void poll_events(void) {
     SDL_Event e;
 
-    ctx.game.window_size_has_changed = false;
+    ctx.window_size_has_changed = false;
 
     while (SDL_PollEvent(&e)) {
         switch (e.type) {
         case SDL_QUIT:
-            ctx.game.is_running = false;
+            ctx.is_running = false;
             break;
 
         case SDL_WINDOWEVENT:
@@ -69,7 +74,7 @@ static void poll_events(void) {
 
             switch (e.window.event) {
             case SDL_WINDOWEVENT_SIZE_CHANGED:
-                ctx.game.window_size_has_changed = true;
+                ctx.window_size_has_changed = true;
                 break;
 
             default:
@@ -107,6 +112,11 @@ static void APIENTRY opengl_log(GLenum source,
 #endif
 
 
+static void preserve_persistent_ctx_fields(void) {
+    ctx.game.udata = ctx.game_copy.udata;
+}
+
+
 static void main_loop(void) {
     /*
     if (!ctx.is_running) {
@@ -120,7 +130,7 @@ static void main_loop(void) {
     int64_t current_frame_time = SDL_GetPerformanceCounter();
     int64_t delta_time = current_frame_time - ctx.prev_frame_time;
     ctx.prev_frame_time = current_frame_time;
-    ctx.game.delta_time = delta_time;
+    ctx.delta_time = delta_time;
 
     /* handle unexpected timer anomalies (overflow, extra slow frames, etc) */
     if (delta_time > ctx.desired_frametime * 8) { /* ignore extra-slow frames */
@@ -184,24 +194,23 @@ static void main_loop(void) {
         ctx.resync_flag = false;
     }
 
+    ctx.game_copy = ctx.game;
+
     /* finally, let's get to work */
     int frames = 0;
-    while (ctx.frame_accumulator >= ctx.desired_frametime * ctx.game.update_multiplicity) {
+    while (ctx.frame_accumulator >= ctx.desired_frametime * ctx.update_multiplicity) {
         frames += 1;
-        for (size_t i = 0; i < ctx.game.update_multiplicity; ++i) {
+        for (size_t i = 0; i < ctx.update_multiplicity; ++i) {
             /* TODO: disable rendering pushes on not-last ? */
             render_queue_clear();
-
             poll_events();
-
             input_state_update(&ctx.input);
-
             game_object_tick();
+            preserve_persistent_ctx_fields();
 
             ctx.frame_accumulator -= ctx.desired_frametime;
-            ctx.game.tick_count = (ctx.game.tick_count % ULLONG_MAX) + 1;
+            ctx.game.frame_number = (ctx.game.frame_number % ULLONG_MAX) + 1;
             ctx.game.initialization_needed = false;
-
         }
     }
 
@@ -435,7 +444,7 @@ static bool initialize(void) {
             goto fail;
         }
         ctx.base_render_width = datum_base_render_width.u.i;
-        ctx.game.base_draw_w = (int)ctx.base_render_width;
+        ctx.game.resolution.x = (int)ctx.base_render_width;
 
         toml_datum_t datum_base_render_height = toml_int_in(game, "base_render_height");
         if (!datum_base_render_height.ok) {
@@ -443,7 +452,7 @@ static bool initialize(void) {
             goto fail;
         }
         ctx.base_render_height = datum_base_render_height.u.i;
-        ctx.game.base_draw_h = (int)ctx.base_render_height;
+        ctx.game.resolution.y = (int)ctx.base_render_height;
 
         ctx.window = SDL_CreateWindow(datum_title.u.s,
                                       SDL_WINDOWPOS_CENTERED,
@@ -499,8 +508,8 @@ static bool initialize(void) {
 
     /* TODO: */
     // SDL_GetRendererOutputSize(ctx.renderer, &ctx.window_w, &ctx.window_h);
-    ctx.game.window_w = (int)ctx.base_render_width;
-    ctx.game.window_h = (int)ctx.base_render_height;
+    ctx.window_dims.x = (int)ctx.base_render_width;
+    ctx.window_dims.y = (int)ctx.base_render_height;
 
     /* add a watcher for immediate updates on window size */
     SDL_AddEventWatch(event_callback, NULL);
@@ -528,7 +537,7 @@ static bool initialize(void) {
     }
 
     /* you could change this at runtime if you wanted */
-    ctx.game.update_multiplicity = 1;
+    ctx.update_multiplicity = 1;
 
 #ifndef EMSCRIPTEN
     /* hook up opengl debugging callback */
@@ -557,13 +566,13 @@ static bool initialize(void) {
         }
     }
 
-    ctx.game.is_running = true;
+    ctx.is_running = true;
     ctx.resync_flag = true;
     ctx.clocks_per_second = SDL_GetPerformanceFrequency();
     ctx.prev_frame_time = SDL_GetPerformanceCounter();
     ctx.desired_frametime = ctx.clocks_per_second / ctx.ticks_per_second;
     ctx.frame_accumulator = 0;
-    ctx.game.tick_count = 0;
+    ctx.game.frame_number = 0;
 
     /* delta time averaging */
     ctx.delta_averager_residual = 0;
@@ -649,6 +658,8 @@ static bool initialize(void) {
     }
     */
 
+    ctx.render_double_buffered = true;
+
     return true;
 
 fail:
@@ -768,7 +779,7 @@ int enter_loop(int argc, char **argv) {
     ctx.was_successful = true;
     ctx.game.initialization_needed = true;
 
-    while (ctx.game.is_running) {
+    while (ctx.is_running) {
         if (game_object_try_reloading()) {
             ctx.game.initialization_needed = true;
             reset_state();

src/twn_stb.c

@@ -0,0 +1,24 @@
+/* single compilation unit for every stb implementation */
+
+#define STB_DS_IMPLEMENTATION
+#define STBDS_ASSERT SDL_assert
+#define STBDS_REALLOC(context,ptr,size) ((void)(context), SDL_realloc(ptr, size))
+#define STBDS_FREE(context,ptr) ((void)(context), SDL_free(ptr))
+#include <stb_ds.h>
+
+#define STB_RECT_PACK_IMPLEMENTATION
+#define STBRP_SORT SDL_qsort
+#define STBRP_ASSERT SDL_assert
+#include <stb_rect_pack.h>
+
+#define STB_TRUETYPE_IMPLEMENTATION
+#define STBTT_malloc(x,u) ((void)(u), SDL_malloc(x))
+#define STBTT_free(x,u) ((void)(u), SDL_free(x))
+#define STBTT_assert(x) SDL_assert(x)
+#define STBTT_strlen(x) SDL_strlen(x)
+#define STBTT_memcpy SDL_memcpy
+#define STBTT_memset SDL_memset
+#include <stb_truetype.h>
+
+#define STB_IMAGE_IMPLEMENTATION
+#include <stb_image.h>

src/twn_textures.c

@@ -1,5 +1,4 @@
 #include "twn_textures_c.h"
-#include "twn_config.h"
 #include "twn_util.h"
 #include "twn_util_c.h"
 #include "twn_engine_context_c.h"
@@ -9,8 +8,6 @@
 #include <physfsrwops.h>
 #include <stb_ds.h>
 #include <stb_rect_pack.h>
-
-#define STB_IMAGE_IMPLEMENTATION
 #include <stb_image.h>
 
 #include <stdbool.h>
@@ -397,7 +394,7 @@ void textures_update_atlas(TextureCache *cache) {
             add_new_atlas(cache);
             ++cache->atlas_index;
         }
-    };
+    }
 
     update_texture_rects_in_atlas(cache, rects);
     recreate_current_atlas_texture(cache);

src/twn_textures_c.h

@@ -11,6 +11,11 @@
 #include <stdbool.h>
 
 
+#define TEXTURE_ATLAS_SIZE_DEFAULT 2048
+#define TEXTURE_ATLAS_BIT_DEPTH 32
+#define TEXTURE_ATLAS_FORMAT SDL_PIXELFORMAT_RGBA32
+
+
 typedef struct Texture {
     Rect srcrect;            /* position in atlas */
     SDL_Surface *data;          /* original image data */
@@ -45,6 +50,7 @@ typedef struct TextureCache {
 typedef struct TextureKey { uint16_t id; } TextureKey;
 
 /* tests whether given key structure corresponds to any texture */
+#define TEXTURE_KEY_INVALID (TextureKey) { (uint16_t)-1 }
 #define m_texture_key_is_valid(p_key) ((p_key).id != (uint16_t)-1)
 
 void textures_cache_init(struct TextureCache *cache, SDL_Window *window);

src/twn_util.c

@@ -4,23 +4,7 @@
 
 #include <SDL2/SDL.h>
 #include <physfsrwops.h>
-#define STB_DS_IMPLEMENTATION
-#define STBDS_ASSERT SDL_assert
-#define STBDS_REALLOC(context,ptr,size) ((void)(context), SDL_realloc(ptr, size))
-#define STBDS_FREE(context,ptr) ((void)(context), SDL_free(ptr))
 #include <stb_ds.h>
-#define STB_RECT_PACK_IMPLEMENTATION
-#define STBRP_SORT SDL_qsort
-#define STBRP_ASSERT SDL_assert
-#include <stb_rect_pack.h>
-#define STB_TRUETYPE_IMPLEMENTATION
-#define STBTT_malloc(x,u) ((void)(u), SDL_malloc(x))
-#define STBTT_free(x,u) ((void)(u), SDL_free(x))
-#define STBTT_assert(x) SDL_assert(x)
-#define STBTT_strlen(x) SDL_strlen(x)
-#define STBTT_memcpy SDL_memcpy
-#define STBTT_memset SDL_memset
-#include <stb_truetype.h>
 
 #include <stdarg.h>
 
@@ -33,10 +17,16 @@ void cry_impl(const char *file, const int line, const char *title, const char *t
 
 
 static void log_impl(const char *restrict format, va_list args, SDL_LogPriority priority) {
+
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wformat-nonliteral"
+
     SDL_LogMessageV(SDL_LOG_CATEGORY_APPLICATION,
                     priority,
                     format,
                     args);
+
+#pragma GCC diagnostic pop
 }
 
 
@@ -125,7 +115,6 @@ int clampi(int i, int min, int max) {
 }
 
 
-
 int64_t file_to_bytes(const char *path, unsigned char **buf_out) {
     SDL_RWops *handle = PHYSFSRWOPS_openRead(path);
 
@@ -272,12 +261,14 @@ void tick_timer(int *value) {
     *value = MAX(*value - 1, 0);
 }
 
+
 void tick_ftimer(float *value) {
-    *value = MAX(*value - ((float)ctx.game.delta_time / (float)ctx.clocks_per_second), 0.0f);
+    *value = MAX(*value - ((float)ctx.delta_time / (float)ctx.clocks_per_second), 0.0f);
 }
 
+
 bool repeat_ftimer(float *value, float at) {
-    *value -= (float)ctx.game.delta_time / (float)ctx.clocks_per_second;
+    *value -= (float)ctx.delta_time / (float)ctx.clocks_per_second;
     if (*value < 0.0f) {
         *value += at;
         return true;

src/twn_util_c.h

@@ -3,6 +3,23 @@
 
 #include "twn_types.h"
 
+#include <SDL2/SDL.h>
+
+#include <stdbool.h>
+
+
+#define MAX SDL_max
+#define MIN SDL_min
+
+void cry_impl(const char *file, const int line, const char *title, const char *text);
+#define CRY(title, text) cry_impl(__FILE__, __LINE__, title, text)
+#define CRY_SDL(title) cry_impl(__FILE__, __LINE__, title, SDL_GetError())
+#define CRY_PHYSFS(title)                                               \
+cry_impl(__FILE__, __LINE__, title, PHYSFS_getErrorByCode(PHYSFS_getLastErrorCode()))
+
+/* for when there's absolutely no way to continue */
+_Noreturn void die_abruptly(void);
+
 /* note: you must free the returned string */
 char *expand_asterisk(const char *mask, const char *to);
 

third-party/physfs/CMakeLists.txt

@@ -85,16 +85,7 @@ set(PHYSFS_SRCS
     src/physfs_platform_android.c
     src/physfs_archiver_dir.c
     src/physfs_archiver_unpacked.c
-    src/physfs_archiver_grp.c
-    src/physfs_archiver_hog.c
-    src/physfs_archiver_7z.c
-    src/physfs_archiver_mvl.c
-    src/physfs_archiver_qpak.c
-    src/physfs_archiver_wad.c
     src/physfs_archiver_zip.c
-    src/physfs_archiver_slb.c
-    src/physfs_archiver_iso9660.c
-    src/physfs_archiver_vdf.c
     ${PHYSFS_CPP_SRCS}
     ${PHYSFS_M_SRCS}
 )

third-party/stb/stb_ds.h

@@ -476,7 +476,7 @@ extern void stbds_rand_seed(size_t seed);
 
 // these are the hash functions used internally if you want to test them or use them for other purposes
 extern size_t stbds_hash_bytes(void *p, size_t len, size_t seed);
-extern size_t stbds_hash_string(char *str, size_t seed);
+extern size_t stbds_hash_string(char const*str, size_t seed);
 
 // this is a simple string arena allocator, initialize with e.g. 'stbds_string_arena my_arena={0}'.
 typedef struct stbds_string_arena stbds_string_arena;
@@ -494,11 +494,11 @@ extern void stbds_unit_tests(void);
 extern void * stbds_arrgrowf(void *a, size_t elemsize, size_t addlen, size_t min_cap);
 extern void   stbds_arrfreef(void *a);
 extern void   stbds_hmfree_func(void *p, size_t elemsize);
-extern void * stbds_hmget_key(void *a, size_t elemsize, void *key, size_t keysize, int mode);
-extern void * stbds_hmget_key_ts(void *a, size_t elemsize, void *key, size_t keysize, ptrdiff_t *temp, int mode);
+extern void * stbds_hmget_key(void *a, size_t elemsize, void const*key, size_t keysize, int mode);
+extern void * stbds_hmget_key_ts(void *a, size_t elemsize, void const*key, size_t keysize, ptrdiff_t *temp, int mode);
 extern void * stbds_hmput_default(void *a, size_t elemsize);
-extern void * stbds_hmput_key(void *a, size_t elemsize, void *key, size_t keysize, int mode);
-extern void * stbds_hmdel_key(void *a, size_t elemsize, void *key, size_t keysize, size_t keyoffset, int mode);
+extern void * stbds_hmput_key(void *a, size_t elemsize, void const*key, size_t keysize, int mode);
+extern void * stbds_hmdel_key(void *a, size_t elemsize, void const*key, size_t keysize, size_t keyoffset, int mode);
 extern void * stbds_shmode_func(size_t elemsize, int mode);
 
 #ifdef __cplusplus
@@ -531,7 +531,7 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
 
 #define STBDS_OFFSETOF(var,field)           ((char *) &(var)->field - (char *) (var))
 
-#define stbds_header(t)  ((stbds_array_header *) (t) - 1)
+#define stbds_header(t)  (((stbds_array_header *) (void *) (t)) - 1)
 #define stbds_temp(t)    stbds_header(t)->temp
 #define stbds_temp_key(t) (*(char **) stbds_header(t)->hash_table)
 
@@ -561,7 +561,7 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
 #define stbds_arrgrow(a,b,c)   ((a) = stbds_arrgrowf_wrapper((a), sizeof *(a), (b), (c)))
 
 #define stbds_hmput(t, k, v) \
-    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, 0),   \
+    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void const*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, 0),   \
      (t)[stbds_temp((t)-1)].key = (k),    \
      (t)[stbds_temp((t)-1)].value = (v))
 
@@ -570,11 +570,11 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
      (t)[stbds_temp((t)-1)] = (s))
 
 #define stbds_hmgeti(t,k) \
-    ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, STBDS_HM_BINARY), \
+    ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void const*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, STBDS_HM_BINARY), \
       stbds_temp((t)-1))
 
 #define stbds_hmgeti_ts(t,k,temp) \
-    ((t) = stbds_hmget_key_ts_wrapper((t), sizeof *(t), (void*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, &(temp), STBDS_HM_BINARY), \
+    ((t) = stbds_hmget_key_ts_wrapper((t), sizeof *(t), (void const*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, &(temp), STBDS_HM_BINARY), \
       (temp))
 
 #define stbds_hmgetp(t, k) \
@@ -584,7 +584,7 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
     ((void) stbds_hmgeti_ts(t,k,temp), &(t)[temp])
 
 #define stbds_hmdel(t,k) \
-    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, STBDS_OFFSETOF((t),key), STBDS_HM_BINARY)),(t)?stbds_temp((t)-1):0)
+    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void const*) STBDS_ADDRESSOF((t)->key, (k)), sizeof (t)->key, STBDS_OFFSETOF((t),key), STBDS_HM_BINARY)),(t)?stbds_temp((t)-1):0)
 
 #define stbds_hmdefault(t, v) \
     ((t) = stbds_hmput_default_wrapper((t), sizeof *(t)), (t)[-1].value = (v))
@@ -603,28 +603,28 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
 #define stbds_hmgetp_null(t,k)  (stbds_hmgeti(t,k) == -1 ? NULL : &(t)[stbds_temp((t)-1)])
 
 #define stbds_shput(t, k, v) \
-    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void*) (k), sizeof (t)->key, STBDS_HM_STRING),   \
+    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void const*) (k), sizeof (t)->key, STBDS_HM_STRING),   \
      (t)[stbds_temp((t)-1)].value = (v))
 
 #define stbds_shputi(t, k, v) \
-    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void*) (k), sizeof (t)->key, STBDS_HM_STRING),   \
+    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void const*) (k), sizeof (t)->key, STBDS_HM_STRING),   \
      (t)[stbds_temp((t)-1)].value = (v), stbds_temp((t)-1))
 
 #define stbds_shputs(t, s) \
-    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void*) (s).key, sizeof (s).key, STBDS_HM_STRING), \
+    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void const*) (s).key, sizeof (s).key, STBDS_HM_STRING), \
      (t)[stbds_temp((t)-1)] = (s), \
      (t)[stbds_temp((t)-1)].key = stbds_temp_key((t)-1)) // above line overwrites whole structure, so must rewrite key here if it was allocated internally
 
 #define stbds_pshput(t, p) \
-    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void*) (p)->key, sizeof (p)->key, STBDS_HM_PTR_TO_STRING), \
+    ((t) = stbds_hmput_key_wrapper((t), sizeof *(t), (void const*) (p)->key, sizeof (p)->key, STBDS_HM_PTR_TO_STRING), \
      (t)[stbds_temp((t)-1)] = (p))
 
 #define stbds_shgeti(t,k) \
-     ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void*) (k), sizeof (t)->key, STBDS_HM_STRING), \
+     ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void const*) (k), sizeof (t)->key, STBDS_HM_STRING), \
       stbds_temp((t)-1))
 
 #define stbds_pshgeti(t,k) \
-     ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void*) (k), sizeof (*(t))->key, STBDS_HM_PTR_TO_STRING), \
+     ((t) = stbds_hmget_key_wrapper((t), sizeof *(t), (void const*) (k), sizeof (*(t))->key, STBDS_HM_PTR_TO_STRING), \
       stbds_temp((t)-1))
 
 #define stbds_shgetp(t, k) \
@@ -634,9 +634,9 @@ extern void * stbds_shmode_func(size_t elemsize, int mode);
     ((void) stbds_pshgeti(t,k), (t)[stbds_temp((t)-1)])
 
 #define stbds_shdel(t,k) \
-    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void*) (k), sizeof (t)->key, STBDS_OFFSETOF((t),key), STBDS_HM_STRING)),(t)?stbds_temp((t)-1):0)
+    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void const*) (k), sizeof (t)->key, STBDS_OFFSETOF((t),key), STBDS_HM_STRING)),(t)?stbds_temp((t)-1):0)
 #define stbds_pshdel(t,k) \
-    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void*) (k), sizeof (*(t))->key, STBDS_OFFSETOF(*(t),key), STBDS_HM_PTR_TO_STRING)),(t)?stbds_temp((t)-1):0)
+    (((t) = stbds_hmdel_key_wrapper((t),sizeof *(t), (void const*) (k), sizeof (*(t))->key, STBDS_OFFSETOF(*(t),key), STBDS_HM_PTR_TO_STRING)),(t)?stbds_temp((t)-1):0)
 
 #define stbds_sh_new_arena(t)  \
     ((t) = stbds_shmode_func_wrapper(t, sizeof *(t), STBDS_SH_ARENA))
@@ -702,10 +702,10 @@ template<class T> static T * stbds_hmget_key_ts_wrapper(T *a, size_t elemsize, v
 template<class T> static T * stbds_hmput_default_wrapper(T *a, size_t elemsize) {
   return (T*)stbds_hmput_default((void *)a, elemsize);
 }
-template<class T> static T * stbds_hmput_key_wrapper(T *a, size_t elemsize, void *key, size_t keysize, int mode) {
+template<class T> static T * stbds_hmput_key_wrapper(T *a, size_t elemsize, void const*key, size_t keysize, int mode) {
   return (T*)stbds_hmput_key((void*)a, elemsize, key, keysize, mode);
 }
-template<class T> static T * stbds_hmdel_key_wrapper(T *a, size_t elemsize, void *key, size_t keysize, size_t keyoffset, int mode){
+template<class T> static T * stbds_hmdel_key_wrapper(T *a, size_t elemsize, void const*key, size_t keysize, size_t keyoffset, int mode){
   return (T*)stbds_hmdel_key((void*)a, elemsize, key, keysize, keyoffset, mode);
 }
 template<class T> static T * stbds_shmode_func_wrapper(T *, size_t elemsize, int mode) {
@@ -1013,7 +1013,7 @@ static stbds_hash_index *stbds_make_hash_index(size_t slot_count, stbds_hash_ind
 #define STBDS_ROTATE_LEFT(val, n)   (((val) << (n)) | ((val) >> (STBDS_SIZE_T_BITS - (n))))
 #define STBDS_ROTATE_RIGHT(val, n)  (((val) >> (n)) | ((val) << (STBDS_SIZE_T_BITS - (n))))
 
-size_t stbds_hash_string(char *str, size_t seed)
+size_t stbds_hash_string(char const*str, size_t seed)
 {
   size_t hash = seed;
   while (*str)
@@ -1278,7 +1278,7 @@ static ptrdiff_t stbds_hm_find_slot(void *a, size_t elemsize, void *key, size_t
   /* NOTREACHED */
 }
 
-void * stbds_hmget_key_ts(void *a, size_t elemsize, void *key, size_t keysize, ptrdiff_t *temp, int mode)
+void * stbds_hmget_key_ts(void *a, size_t elemsize, void const*key, size_t keysize, ptrdiff_t *temp, int mode)
 {
   size_t keyoffset = 0;
   if (a == NULL) {
@@ -1309,7 +1309,7 @@ void * stbds_hmget_key_ts(void *a, size_t elemsize, void *key, size_t keysize, p
   }
 }
 
-void * stbds_hmget_key(void *a, size_t elemsize, void *key, size_t keysize, int mode)
+void * stbds_hmget_key(void *a, size_t elemsize, void const*key, size_t keysize, int mode)
 {
   ptrdiff_t temp;
   void *p = stbds_hmget_key_ts(a, elemsize, key, keysize, &temp, mode);
@@ -1334,7 +1334,7 @@ void * stbds_hmput_default(void *a, size_t elemsize)
 
 static char *stbds_strdup(char *str);
 
-void *stbds_hmput_key(void *a, size_t elemsize, void *key, size_t keysize, int mode)
+void *stbds_hmput_key(void *a, size_t elemsize, void const *key, size_t keysize, int mode)
 {
   size_t keyoffset=0;
   void *raw_a;
@@ -1370,7 +1370,7 @@ void *stbds_hmput_key(void *a, size_t elemsize, void *key, size_t keysize, int m
 
   // we iterate hash table explicitly because we want to track if we saw a tombstone
   {
-    size_t hash = mode >= STBDS_HM_STRING ? stbds_hash_string((char*)key,table->seed) : stbds_hash_bytes(key, keysize,table->seed);
+    size_t hash = mode >= STBDS_HM_STRING ? stbds_hash_string((char const*)key,table->seed) : stbds_hash_bytes(key, keysize,table->seed);
     size_t step = STBDS_BUCKET_LENGTH;
     size_t pos;
     ptrdiff_t tombstone = -1;
@@ -1469,7 +1469,7 @@ void * stbds_shmode_func(size_t elemsize, int mode)
   return STBDS_ARR_TO_HASH(a,elemsize);
 }
 
-void * stbds_hmdel_key(void *a, size_t elemsize, void *key, size_t keysize, size_t keyoffset, int mode)
+void * stbds_hmdel_key(void *a, size_t elemsize, void const*key, size_t keysize, size_t keyoffset, int mode)
 {
   if (a == NULL) {
     return 0;