include/twn_types.h

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

src/rendering/twn_circles.c

@@ -1,3 +1,4 @@
+#include "twn_util.h"
 #include "twn_engine_context_c.h"
 #include "twn_draw_c.h"
 #include "twn_draw.h"
@@ -23,15 +24,22 @@ void draw_circle(Vec2 position, float radius, Color color) {
 
 
 void create_circle_geometry(Vec2 position,
+                            Color color,
                             float radius,
                             size_t num_vertices,
-                            Vec2 vertices[])
+                            SDL_Vertex vertices[],
+                            int indices[])
 {
     /* 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].x = (float)position.x;
+    vertices[0].position.x = (float)position.x;
-    vertices[0].y = (float)position.y;
+    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 };
 
     /* this point will rotate around the center */
     float start_x = 0.0f - radius;
@@ -40,13 +48,34 @@ 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;
 
-        float c, s;
+        vertices[i].position.x =
-        sincosf(final_seg_rotation_angle, &s, &c);
+            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;
 
-        vertices[i].x = c * start_x - s * start_y;
+        vertices[i].position.x += position.x;
-        vertices[i].y = c * start_y + s * start_x;
+        vertices[i].position.y += position.y;
 
-        vertices[i].x += position.x;
+        vertices[i].color.r = color.r;
-        vertices[i].y += position.y;
+        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;
     }
 }

src/rendering/twn_draw.c

@@ -374,11 +374,11 @@ void render(void) {
         }
     }
 
-    start_render_frame(); {
     render_space();
-        render_skybox(); /* after space, as to use depth buffer for early z rejection */
+    render_skybox(); /* after space, as to use depth buffer for early rejection */
     render_2d();
-    } end_render_frame();
+    swap_buffers();
+    clear_draw_buffer();
 }
 
 

src/rendering/twn_draw_c.h

@@ -20,10 +20,7 @@
 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;
@@ -132,9 +129,11 @@ 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,
-                            Vec2 vertices[]);
+                            SDL_Vertex vertices[],
+                            int indices[]);
 
 struct QuadBatch {
     size_t size;             /* how many primitives are in current batch */
@@ -166,8 +165,6 @@ 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 const *data, size_t bytes);
@@ -190,9 +187,7 @@ void swap_buffers(void);
 
 void set_depth_range(double low, double high);
 
-VertexBuffer get_quad_element_buffer(void);
+void bind_quad_element_buffer(void);
-
-VertexBuffer get_circle_element_buffer(void);
 
 void render_circle(const CirclePrimitive *circle);
 
@@ -243,8 +238,4 @@ 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_rendering.c

@@ -9,113 +9,92 @@
 #include <stb_ds.h>
 
 
+typedef struct {
+    float x, y;
+} Vec2Packed;
+
+
+typedef struct {
+    int16_t x, y;
+} Vec2ShortPacked;
+
+
+typedef struct {
+    uint8_t r, g, b, a;
+} ColorPacked;
+
+
+#define m_vec2_packed_from(p_vec2) ((Vec2Packed){ (p_vec2).x, (p_vec2).y })
+#define m_vec2_short_packed_from(p_vec2) ((Vec2ShortPacked){ (uint16_t)((p_vec2).x * 32768 * 2), (uint16_t)((p_vec2).y * 32768 * 2) })
+#define m_color_packed_from(p_color) ((ColorPacked){ (p_color).r, (p_color).g, (p_color).b, (p_color).a })
+
+
 /* 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 */
 typedef struct ElementIndexedQuad {
     /* upper-left */
-    Vec2 v0;
+    Vec2Packed v0;
-    Vec2 uv0;
+    Vec2Packed uv0;
-    Color c0;
+    ColorPacked c0;
     /* bottom-left */
-    Vec2 v1;
+    Vec2Packed v1;
-    Vec2 uv1;
+    Vec2Packed uv1;
-    Color c1;
+    ColorPacked c1;
     /* bottom-right */
-    Vec2 v2;
+    Vec2Packed v2;
-    Vec2 uv2;
+    Vec2Packed uv2;
-    Color c2;
+    ColorPacked c2;
     /* upper-right */
-    Vec2 v3;
+    Vec2Packed v3;
-    Vec2 uv3;
+    Vec2Packed uv3;
-    Color c3;
+    ColorPacked c3;
 } ElementIndexedQuad;
 
 typedef struct ElementIndexedQuadWithoutColor {
     /* upper-left */
-    Vec2 v0;
+    Vec2Packed v0;
-    Vec2 uv0;
+    Vec2Packed uv0;
     /* bottom-left */
-    Vec2 v1;
+    Vec2Packed v1;
-    Vec2 uv1;
+    Vec2Packed uv1;
     /* bottom-right */
-    Vec2 v2;
+    Vec2Packed v2;
-    Vec2 uv2;
+    Vec2Packed uv2;
     /* upper-right */
-    Vec2 v3;
+    Vec2Packed v3;
-    Vec2 uv3;
+    Vec2Packed uv3;
 } ElementIndexedQuadWithoutColor;
 
 
 typedef struct ElementIndexedQuadWithoutTexture {
     /* upper-left */
-    Vec2 v0;
+    Vec2Packed v0;
-    Color c0;
+    ColorPacked c0;
     /* bottom-left */
-    Vec2 v1;
+    Vec2Packed v1;
-    Color c1;
+    ColorPacked c1;
     /* bottom-right */
-    Vec2 v2;
+    Vec2Packed v2;
-    Color c2;
+    ColorPacked c2;
     /* upper-right */
-    Vec2 v3;
+    Vec2Packed v3;
-    Color c3;
+    ColorPacked c3;
 } ElementIndexedQuadWithoutTexture;
 
 
 typedef struct ElementIndexedQuadWithoutColorWithoutTexture {
     /* upper-left */
-    Vec2 v0;
+    Vec2Packed v0;
     /* bottom-left */
-    Vec2 v1;
+    Vec2Packed v1;
     /* bottom-right */
-    Vec2 v2;
+    Vec2Packed v2;
     /* upper-right */
-    Vec2 v3;
+    Vec2Packed 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,
@@ -123,266 +102,13 @@ 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 */
+#define CIRCLE_VERTICES_MAX 2048
-/* 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;
 
@@ -415,22 +141,11 @@ static void finally_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();
@@ -467,25 +182,45 @@ static void finally_use_2d_pipeline(void) {
     glMatrixMode(GL_MODELVIEW);
     glLoadIdentity();
 
-    texture_mode_last_used = -1;
     pipeline_last_used = PIPELINE_2D;
 }
 
 
-void use_texture_mode(TextureMode mode) {
+void render_circle(const CirclePrimitive *circle) {
-    DeferredCommand const command = {
+    static SDL_Vertex vertices[CIRCLE_VERTICES_MAX];
-        .type = DEFERRED_COMMAND_TYPE_USE_TEXTURE_MODE,
+    static int indices[CIRCLE_VERTICES_MAX * 3];
-        .use_texture_mode = { mode }
+    int num_vertices = MIN((int)circle->radius, CIRCLE_VERTICES_MAX-1);
-    };
 
-    arrpush(deferred_commands, command);
+    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);
 }
 
 
-static void finally_use_texture_mode(TextureMode mode) {
+void use_texture_mode(TextureMode mode) {
-    if (texture_mode_last_used == mode)
-        return;
-
     static GLuint lists = 0;
     if (!lists) {
         lists = glGenLists(3);
@@ -524,8 +259,6 @@ static void finally_use_texture_mode(TextureMode mode) {
     } else {
         glCallList(lists + 2);
     }
-
-    texture_mode_last_used = mode;
 }
 
 
@@ -549,7 +282,7 @@ bool push_to_vertex_buffer_builder(VertexBufferBuilder *builder,
     if (builder->bytes_left == 0)
         return false;
 
-    memcpy(builder->mapping, bytes, size);
+    SDL_memcpy(builder->mapping, bytes, size);
     builder->bytes_left -= size;
 
     /* trigger data send */
@@ -568,7 +301,7 @@ void finally_render_quads(const Primitive2D primitives[],
                           const struct QuadBatch batch,
                           const VertexBuffer buffer)
 {
-    DeferredCommandDraw command = {0};
+    (void)buffer;
 
     GLsizei off = 0, voff = 0, uvoff = 0, coff = 0;
 
@@ -590,55 +323,63 @@ void finally_render_quads(const Primitive2D primitives[],
         voff  = offsetof(ElementIndexedQuadWithoutColorWithoutTexture, v0);
     }
 
-    command.vertices = (AttributeArrayPointer) {
+    /* vertex specification */
-        .arity = 2,
+    glEnableClientState(GL_VERTEX_ARRAY);
-        .type = GL_FLOAT,
+    glVertexPointer(2,
-        .stride = off,
+                    GL_FLOAT,
-        .offset = voff,
+                    off,
-        .buffer = buffer
+                    (void *)(size_t)voff);
-    };
-
-    if (batch.textured)
-        command.texcoords = (AttributeArrayPointer) {
-            .arity = 2,
-            .type = GL_FLOAT,
-            .stride = off,
-            .offset = uvoff,
-            .buffer = buffer
-        };
-
-    if (!batch.constant_colored) {
-        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.textured) {
-        /* TODO: bad, don't */
+        glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-        command.textured = true;
+        glClientActiveTexture(GL_TEXTURE0);
-        command.texture_key = primitives->sprite.texture_key;
+        glTexCoordPointer(2,
-        command.texture_repeat = batch.repeat;
+                          GL_FLOAT,
+                          off,
+                          (void *)(size_t)uvoff);
     }
 
-    command.element_buffer = get_quad_element_buffer();
+    if (!batch.constant_colored) {
-    command.element_count = 6 * (GLsizei)batch.size;
+        glEnableClientState(GL_COLOR_ARRAY);
-    command.range_end = 6 * (GLsizei)batch.size;
+        glColorPointer(4,
+                       GL_UNSIGNED_BYTE,
+                       off,
+                       (void *)(size_t)coff);
+    } else
+        glColor4ub(primitives[0].sprite.color.r,
+                   primitives[0].sprite.color.g,
+                   primitives[0].sprite.color.b,
+                   primitives[0].sprite.color.a);
 
-    use_texture_mode(batch.mode);
+    if (batch.textured) {
+        if (!batch.repeat)
+            textures_bind(&ctx.texture_cache, primitives->sprite.texture_key);
+        else
+            textures_bind_repeating(&ctx.texture_cache, primitives->sprite.texture_key);
+    }
 
-    DeferredCommand final_command = {
+    bind_quad_element_buffer();
-        .type = DEFERRED_COMMAND_TYPE_DRAW,
-        .draw = command
-    };
 
-    arrpush(deferred_commands, final_command);
+    glDrawElements(GL_TRIANGLES, 6 * (GLsizei)batch.size, GL_UNSIGNED_SHORT, NULL);
+
+    /* clear the state */
+    {
+        glBufferData(GL_ARRAY_BUFFER, 0, NULL, GL_STREAM_DRAW);
+        glBindBuffer(GL_ARRAY_BUFFER, 0);
+
+        if (batch.textured)
+            glDisableClientState(GL_TEXTURE_COORD_ARRAY);
+
+        glDisableClientState(GL_VERTEX_ARRAY);
+
+        if (!batch.constant_colored)
+            glDisableClientState(GL_COLOR_ARRAY);
+
+        if (batch.textured)
+            glBindTexture(GL_TEXTURE_2D, 0);
+
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
+    }
 }
 
 
@@ -665,62 +406,62 @@ bool push_quad_payload_to_vertex_buffer_builder(struct QuadBatch batch,
 {
     if (!batch.constant_colored && batch.textured) {
         ElementIndexedQuad const buffer_element = {
-            .v0 = v0,
+            .v0 = m_vec2_packed_from(v0),
-            .v1 = v1,
+            .v1 = m_vec2_packed_from(v1),
-            .v2 = v2,
+            .v2 = m_vec2_packed_from(v2),
-            .v3 = v3,
+            .v3 = m_vec2_packed_from(v3),
 
-            .uv0 = uv0,
+            .uv0 = m_vec2_packed_from(uv0),
-            .uv1 = uv1,
+            .uv1 = m_vec2_packed_from(uv1),
-            .uv2 = uv2,
+            .uv2 = m_vec2_packed_from(uv2),
-            .uv3 = uv3,
+            .uv3 = m_vec2_packed_from(uv3),
 
             /* equal for all (flat shaded) */
-            .c0 = color,
+            .c0 = m_color_packed_from(color),
-            // .c1 = color,
+            // .c1 = m_color_packed_from(color),
-            .c2 = color,
+            .c2 = m_color_packed_from(color),
-            // .c3 = color,
+            // .c3 = m_color_packed_from(color),
         };
 
         return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
 
     } else if (batch.constant_colored && batch.textured) {
         ElementIndexedQuadWithoutColor const buffer_element = {
-            .v0 = v0,
+            .v0 = m_vec2_packed_from(v0),
-            .v1 = v1,
+            .v1 = m_vec2_packed_from(v1),
-            .v2 = v2,
+            .v2 = m_vec2_packed_from(v2),
-            .v3 = v3,
+            .v3 = m_vec2_packed_from(v3),
 
-            .uv0 = uv0,
+            .uv0 = m_vec2_packed_from(uv0),
-            .uv1 = uv1,
+            .uv1 = m_vec2_packed_from(uv1),
-            .uv2 = uv2,
+            .uv2 = m_vec2_packed_from(uv2),
-            .uv3 = uv3,
+            .uv3 = m_vec2_packed_from(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,
+            .v0 = m_vec2_packed_from(v0),
-            .v1 = v1,
+            .v1 = m_vec2_packed_from(v1),
-            .v2 = v2,
+            .v2 = m_vec2_packed_from(v2),
-            .v3 = v3,
+            .v3 = m_vec2_packed_from(v3),
 
             /* equal for all (flat shaded) */
-            .c0 = color,
+            .c0 = m_color_packed_from(color),
-            // .c1 = color,
+            // .c1 = m_color_packed_from(color),
-            .c2 = color,
+            .c2 = m_color_packed_from(color),
-            // .c3 = color,
+            // .c3 = m_color_packed_from(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,
+            .v0 = m_vec2_packed_from(v0),
-            .v1 = v1,
+            .v1 = m_vec2_packed_from(v1),
-            .v2 = v2,
+            .v2 = m_vec2_packed_from(v2),
-            .v3 = v3,
+            .v3 = m_vec2_packed_from(v3),
         };
 
         return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
@@ -776,15 +517,15 @@ bool push_text_payload_to_vertex_buffer_builder(FontData const *font_data,
     (void)font_data;
 
     ElementIndexedQuadWithoutColor buffer_element = {
-        .v0 = (Vec2){ quad.x0, quad.y0 },
+        .v0 = (Vec2Packed){ quad.x0, quad.y0 },
-        .v1 = (Vec2){ quad.x1, quad.y0 },
+        .v1 = (Vec2Packed){ quad.x1, quad.y0 },
-        .v2 = (Vec2){ quad.x1, quad.y1 },
+        .v2 = (Vec2Packed){ quad.x1, quad.y1 },
-        .v3 = (Vec2){ quad.x0, quad.y1 },
+        .v3 = (Vec2Packed){ quad.x0, quad.y1 },
 
-        .uv0 = (Vec2){ quad.s0, quad.t0 },
+        .uv0 = (Vec2Packed){ quad.s0, quad.t0 },
-        .uv1 = (Vec2){ quad.s1, quad.t0 },
+        .uv1 = (Vec2Packed){ quad.s1, quad.t0 },
-        .uv2 = (Vec2){ quad.s1, quad.t1 },
+        .uv2 = (Vec2Packed){ quad.s1, quad.t1 },
-        .uv3 = (Vec2){ quad.s0, quad.t1 },
+        .uv3 = (Vec2Packed){ quad.s0, quad.t1 },
     };
 
     return push_to_vertex_buffer_builder(builder, &buffer_element, sizeof buffer_element);
@@ -796,46 +537,44 @@ void finally_draw_text(FontData const *font_data,
                        Color color,
                        VertexBuffer buffer)
 {
-    DeferredCommandDraw command = {0};
+    (void)buffer;
 
-    command.vertices = (AttributeArrayPointer) {
+    /* vertex specification */
-        .arity = 2,
+    glEnableClientState(GL_VERTEX_ARRAY);
-        .type = GL_FLOAT,
+    glVertexPointer(2,
-        .stride = offsetof(ElementIndexedQuadWithoutColor, v1),
+                    GL_FLOAT,
-        .offset = offsetof(ElementIndexedQuadWithoutColor, v0),
+                    offsetof(ElementIndexedQuadWithoutColor, v1),
-        .buffer = buffer
+                    (void *)(size_t)offsetof(ElementIndexedQuadWithoutColor, v0));
-    };
 
-    command.texcoords = (AttributeArrayPointer) {
+    glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-        .arity = 2,
+    glClientActiveTexture(GL_TEXTURE0);
-        .type = GL_FLOAT,
+    glTexCoordPointer(2,
-        .stride = offsetof(ElementIndexedQuadWithoutColor, v1),
+                      GL_FLOAT,
-        .offset = offsetof(ElementIndexedQuadWithoutColor, uv0),
+                      offsetof(ElementIndexedQuadWithoutColor, v1),
-        .buffer = buffer
+                      (void *)(size_t)offsetof(ElementIndexedQuadWithoutColor, uv0));
-    };
 
-    command.constant_colored = true;
+    bind_quad_element_buffer();
-    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);
 
-    DeferredCommand final_command = {
+    glBindTexture(GL_TEXTURE_2D, font_data->texture);
-        .type = DEFERRED_COMMAND_TYPE_DRAW,
-        .draw = command
-    };
 
-    arrpush(deferred_commands, final_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);
 
     /* TODO: why doesn't it get restored if not placed here? */
-    // glDepthMask(GL_TRUE);
+    glDepthMask(GL_TRUE);
 }
 
 
@@ -864,47 +603,6 @@ 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;
@@ -1071,16 +769,3 @@ 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

@@ -32,60 +32,65 @@ void specify_vertex_buffer(VertexBuffer buffer, void const *data, size_t bytes)
 }
 
 
-VertexBuffer get_quad_element_buffer(void) {
+void bind_quad_element_buffer(void) {
-    static VertexBuffer buffer = 0;
+    static GLuint buffer = 0;
 
     /* it's only generated once at runtime */
-    /* TODO: use builder interface, not direct calls (glMapBuffer isn't portable) */
     if (buffer == 0) {
-        buffer = create_vertex_buffer();
+        glGenBuffers(1, &buffer);
-        VertexBufferBuilder builder = build_vertex_buffer(buffer, sizeof (GLshort) * QUAD_ELEMENT_BUFFER_LENGTH * 6 );
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer);
+        glBufferData(GL_ELEMENT_ARRAY_BUFFER,
+                     QUAD_ELEMENT_BUFFER_LENGTH * 6 * sizeof(uint16_t),
+                     NULL,
+                     GL_STATIC_DRAW);
 
-        for (size_t i = 0; i < QUAD_ELEMENT_BUFFER_LENGTH; ++i) {
+        uint16_t *const indices = glMapBuffer(GL_ELEMENT_ARRAY_BUFFER,
-            GLshort indices[6];
+                                              GL_WRITE_ONLY);
-            indices[0] = (GLshort)(i * 4 + 0);
+        if (!indices)
-            indices[1] = (GLshort)(i * 4 + 1);
+            CRY("Quad indices generation", "glMapBuffer() failed");
-            indices[2] = (GLshort)(i * 4 + 2);
-            indices[3] = (GLshort)(i * 4 + 2);
-            indices[4] = (GLshort)(i * 4 + 3);
-            indices[5] = (GLshort)(i * 4 + 0);
 
-            push_to_vertex_buffer_builder(&builder, indices, sizeof indices);
+        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);
         }
 
-    SDL_assert_always(buffer);
+        glUnmapBuffer(GL_ELEMENT_ARRAY_BUFFER);
 
-    return buffer;
+    } else
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer);
 }
 
 
-VertexBuffer get_circle_element_buffer(void) {
+void clear_draw_buffer(void) {
-    static VertexBuffer buffer = 0;
+    /* 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);
 
-    if (buffer == 0) {
+    glDepthRange(0.0, 1.0);
-        buffer = create_vertex_buffer();
+    glDepthMask(GL_TRUE);
-        VertexBufferBuilder builder = build_vertex_buffer(buffer, sizeof (GLshort) * CIRCLE_ELEMENT_BUFFER_LENGTH);
 
-        for (size_t i = 1; i < CIRCLE_VERTICES_MAX; ++i) {
+    /* TODO: don't clear color when skybox is applied? */
-            /* first one is center point index, always zero */
+    /*       for that window should match framebuffer */
-            GLshort indices[3];
+    /*       also it is driver dependent, from what i can gather */
-
+    /* INFO: also, based on below, driver might prefer it staying this way */
-            indices[0] = 0;
+    /*       https://gamedev.stackexchange.com/questions/90344/render-with-const-depth-value */
-
+    /*       we could optionally load ARB_invalidate_subdata extension if it's available instead */
-            /* generated point index */
+    glClear(GL_COLOR_BUFFER_BIT |
-            indices[1] = (GLshort)i;
+            GL_DEPTH_BUFFER_BIT |
-
+            GL_STENCIL_BUFFER_BIT);
-            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;
+void swap_buffers(void) {
-
+    SDL_GL_SwapWindow(ctx.window);
-            push_to_vertex_buffer_builder(&builder, indices, sizeof indices);
+}
-        }
+
-    }
+
-
+void set_depth_range(double low, double high) {
-    SDL_assert_always(buffer);
+    glDepthRange(low, high);
-
-    return buffer;
 }

src/rendering/twn_rects.c

@@ -72,7 +72,9 @@ void render_rect_batch(const Primitive2D primitives[],
     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();
+    static VertexBuffer vertex_array = 0;
+    if (vertex_array == 0)
+        vertex_array = create_vertex_buffer();
 
     use_texture_mode(batch.mode);
 

src/rendering/twn_sprites.c

@@ -125,15 +125,16 @@ void render_sprite_batch(const Primitive2D primitives[],
     SDL_assert(primitives && batch.size != 0);
     SDL_assert(primitives[0].type == PRIMITIVE_2D_SPRITE);
 
-    VertexBuffer const vertex_array = get_scratch_vertex_array();
+    /* 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();
+
+    use_texture_mode(batch.mode);
 
     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_quad_payload_size(batch) * batch.size);
@@ -143,16 +144,13 @@ void render_sprite_batch(const Primitive2D primitives[],
             const size_t cur = batch.mode == TEXTURE_MODE_GHOSTLY ? i : batch.size - i - 1;
             const SpritePrimitive sprite = primitives[cur].sprite;
 
-            if (primitives[cur].sprite.texture_key.id != cached_srcrect_key.id) {
+            /* TODO: try caching it */
-                cached_srcrect = textures_get_srcrect(&ctx.texture_cache, primitives[cur].sprite.texture_key);
+            const Rect srcrect =
-                cached_srcrect_key = primitives[cur].sprite.texture_key;
+                textures_get_srcrect(&ctx.texture_cache, primitives[cur].sprite.texture_key);
-            }
-
-            Rect const srcrect = cached_srcrect;
 
             Vec2 uv0, uv1, uv2, uv3;
 
-            if (!batch.repeat) {
+            if (!sprite.repeat) {
                 const float wr = srcrect.w / dims.w;
                 const float hr = srcrect.h / dims.h;
                 const float xr = srcrect.x / dims.w;

src/rendering/twn_text.c

@@ -171,7 +171,9 @@ static void text_destroy_font_data(FontData *font_data) {
 
 
 static void text_draw_with(FontData* font_data, char* text, Vec2 position, Color color) {
-    VertexBuffer const vertex_array = get_scratch_vertex_array();
+    VertexBuffer vertex_array = 0;
+    if (vertex_array == 0)
+        vertex_array = create_vertex_buffer();
 
     const size_t len = SDL_strlen(text);
 

src/rendering/twn_triangles.c

@@ -45,7 +45,9 @@ void draw_triangle(const char *path,
 void draw_uncolored_space_traingle_batch(struct MeshBatch *batch,
                                          TextureKey texture_key)
 {
-    VertexBuffer const vertex_array = get_scratch_vertex_array();
+    static VertexBuffer vertex_array = 0;
+    if (vertex_array == 0)
+        vertex_array = create_vertex_buffer();
 
     const size_t primitives_len = arrlenu(batch->primitives);
 

src/twn_engine_context_c.h

@@ -77,7 +77,6 @@ typedef struct EngineContext {
     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_loop.c

@@ -658,8 +658,6 @@ static bool initialize(void) {
     }
     */
 
-    ctx.render_double_buffered = true;
-
     return true;
 
 fail:

src/twn_textures_c.h

@@ -50,7 +50,6 @@ 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.h

@@ -39,7 +39,7 @@ static inline float fast_sqrt(float x)
 
 
 static inline Vec2 fast_cossine(float a) {
-    const float s = sinf(a);
+    const float s = SDL_sinf(a);
     return (Vec2){
         .x = fast_sqrt(1.0f - s * s) * (a >= (float)M_PI_2 && a < (float)(M_PI + M_PI_2) ? -1 : 1),
         .y = s