diff --git a/build.bat b/build.bat
index 0d36a88..48f703a 100644
--- a/build.bat
+++ b/build.bat
@@ -5,5 +5,5 @@ set gles_shaders=src\gles_canvas_shaders\gles_canvas_blit_vertex.glsl src\gles_c
 call python scripts\embed_text.py %gles_shaders% --output src\gles_canvas_shaders.h
 
 set INCLUDES=/I src /I src/util /I src/platform /I ext /I ext/angle_headers
-cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GLES /std:c11 %INCLUDES% /c /Fo:bin/milepost.obj src/milepost.c
+cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GL /std:c11 %INCLUDES% /c /Fo:bin/milepost.obj src/milepost.c
 lib bin/milepost.obj /OUT:bin/milepost.lib
diff --git a/examples/perf_text/build.bat b/examples/perf_text/build.bat
index f3cdbf2..beebece 100644
--- a/examples/perf_text/build.bat
+++ b/examples/perf_text/build.bat
@@ -1,4 +1,4 @@
 
 set INCLUDES=/I ..\..\src /I ..\..\src\util /I ..\..\src\platform /I ../../ext /I ../../ext/angle_headers
 
-cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GLES /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib /LIBPATH:../../bin libEGL.dll.lib libGLESv2.dll.lib user32.lib opengl32.lib gdi32.lib /out:../../bin/perf_text.exe
+cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GL /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib user32.lib opengl32.lib gdi32.lib /out:../../bin/perf_text.exe
diff --git a/examples/perf_text/main.c b/examples/perf_text/main.c
index 8daa26b..ec16b43 100644
--- a/examples/perf_text/main.c
+++ b/examples/perf_text/main.c
@@ -86,7 +86,7 @@ int main()
 #if defined(OS_MACOS)
 	mg_surface surface = mg_metal_surface_create_for_window(window);
 #elif defined(OS_WIN64)
-	mg_surface surface = mg_gles_surface_create_for_window(window);
+	mg_surface surface = mg_gl_surface_create_for_window(window);
 #else
 	#error "unsupported OS"
 #endif
diff --git a/examples/triangleGL/build.bat b/examples/triangleGL/build.bat
index 3e918f6..f0aee81 100644
--- a/examples/triangleGL/build.bat
+++ b/examples/triangleGL/build.bat
@@ -1,2 +1,2 @@
-set INCLUDES=/I ..\..\src /I ..\..\src\util /I ..\..\src\platform /I ../../ext
-cl /we4013 /Zi /Zc:preprocessor /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib user32.lib opengl32.lib gdi32.lib /out:test.exe
+set INCLUDES=/I ..\..\src /I ..\..\src\util /I ..\..\src\platform /I ../../ext
+cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GL /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib user32.lib opengl32.lib gdi32.lib /out:test.exe
diff --git a/examples/triangleGL/main.c b/examples/triangleGL/main.c
index 7139f1b..231318f 100644
--- a/examples/triangleGL/main.c
+++ b/examples/triangleGL/main.c
@@ -13,7 +13,6 @@
 #include<math.h>
 
 #include"milepost.h"
-#include"win32_gl_surface.h"
 
 #define LOG_SUBSYSTEM "Main"
 
diff --git a/examples/win_canvas/build.bat b/examples/win_canvas/build.bat
index bf1053a..8a4c8c1 100644
--- a/examples/win_canvas/build.bat
+++ b/examples/win_canvas/build.bat
@@ -1,3 +1,3 @@
 
 set INCLUDES=/I ..\..\src /I ..\..\src\util /I ..\..\src\platform /I ../../ext /I ../../ext/angle_headers
-cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GLES /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib /LIBPATH:../../bin libEGL.dll.lib libGLESv2.dll.lib user32.lib opengl32.lib gdi32.lib /out:../../bin/example_canvas.exe
+cl /we4013 /Zi /Zc:preprocessor /DMG_IMPLEMENTS_BACKEND_GL /std:c11 %INCLUDES% main.c /link /LIBPATH:../../bin milepost.lib user32.lib opengl32.lib gdi32.lib /out:../../bin/example_canvas.exe
diff --git a/examples/win_canvas/main.c b/examples/win_canvas/main.c
index 27c693d..202776f 100644
--- a/examples/win_canvas/main.c
+++ b/examples/win_canvas/main.c
@@ -65,7 +65,7 @@ int main()
 #if defined(OS_MACOS)
 	mg_surface surface = mg_metal_surface_create_for_window(window);
 #elif defined(OS_WIN64)
-	mg_surface surface = mg_gles_surface_create_for_window(window);
+	mg_surface surface = mg_gl_surface_create_for_window(window);
 #else
 	#error "unsupported OS"
 #endif
@@ -158,6 +158,7 @@ int main()
 		y += dy;
 
 		mg_surface_prepare(surface);
+
 			// background
 			mg_set_color_rgba(0, 1, 1, 1);
 			mg_clear();
@@ -196,17 +197,9 @@ int main()
 			mg_fill();
 //*/
 
-//*
 			printf("Milepost vector graphics test program (frame time = %fs, fps = %f)...\n",
 			                      frameTime,
 			                      1./frameTime);
-//*/
-
-/*
-			mg_set_color_rgba(1, 1, 0, 1);
-			mg_rectangle_fill(8, 8, 100, 50);
-*/
-
 
 			mg_flush();
 		mg_surface_present(surface);
diff --git a/ext/angle_install_notes.txt b/ext/angle_install_notes.txt
index 8b52fc5..e6609c2 100644
--- a/ext/angle_install_notes.txt
+++ b/ext/angle_install_notes.txt
@@ -2,8 +2,8 @@ angle install on windows
 
 * need Python3 (can install through win app store)
 * need Windows SDK
-* clone depot_tools git clone https://chromium.googlesource.com/chromium/tools/depot_tools.git 
-or download and unzip bundle at https://storage.googleapis.com/chrome-infra/depot_tools.zip 
+* clone depot_tools git clone https://chromium.googlesource.com/chromium/tools/depot_tools.git
+or download and unzip bundle at https://storage.googleapis.com/chrome-infra/depot_tools.zip
 * set depot_tools in path env variable through control panel>System and security>system>advanced system settings
 * run gclient in a cmd shell
 
@@ -24,8 +24,21 @@ or download and unzip bundle at https://storage.googleapis.com/chrome-infra/depo
 	angle_build_tests = false
 	is_component_build = false
 
+	#to get debugging kinda working with renderdoc:
+	angle_enable_trace = true
+	angle_enable_annotator_run_time_checks = true
+
+
 * autoninja -C out/Debug
 * wait a while
 
 * link with libEGL.dll.lib and libGLESv2.dll.lib
-* put libEGL.dll and libGLESv2.dll in same directory as executable
\ No newline at end of file
+* put libEGL.dll and libGLESv2.dll in same directory as executable
+
+Debugging
+---------
+in renderdoc, set env variables
+RENDERDOC_HOOK_EGL 0 (if you want to trace underlying native API)
+RENDERDOC_HOOK_EGL 1 (if you want to trace EGL calls. You also need to put libEGL in the renderdoc folder so it's found when capturing stuff. Unfortunately though, that seems to provoke crashes...)
+
+ANGLE_ENABLE_DEBUG_MARKERS 1 (to turn on debug markers)
diff --git a/src/gl_canvas.c b/src/gl_canvas.c
new file mode 100644
index 0000000..da18816
--- /dev/null
+++ b/src/gl_canvas.c
@@ -0,0 +1,451 @@
+/************************************************************//**
+*
+*	@file: gl_canvas.c
+*	@author: Martin Fouilleul
+*	@date: 29/01/2023
+*	@revision:
+*
+*****************************************************************/
+#include"graphics_internal.h"
+#include"macro_helpers.h"
+#include"gles_canvas_shaders.h"
+
+#define LOG_SUBSYSTEM "Graphics"
+
+typedef struct mg_gl_canvas_backend
+{
+	mg_canvas_backend interface;
+	mg_surface surface;
+
+	GLint dummyVertexBuffer;
+	GLint vertexBuffer;
+	GLint indexBuffer;
+	GLint tileCounterBuffer;
+	GLint tileArrayBuffer;
+	GLint clearCounterProgram;
+	GLint tileProgram;
+	GLint sortProgram;
+	GLint drawProgram;
+	GLint blitProgram;
+
+	GLint outTexture;
+
+	char* indexMapping;
+	char* vertexMapping;
+
+} mg_gl_canvas_backend;
+
+mg_gl_surface* mg_gl_canvas_get_surface(mg_gl_canvas_backend* canvas)
+{
+	mg_gl_surface* res = 0;
+	mg_surface_data* data = mg_surface_data_from_handle(canvas->surface);
+	if(data && data->backend == MG_BACKEND_GL)
+	{
+		res = (mg_gl_surface*)data;
+	}
+	return(res);
+}
+
+//NOTE: debugger
+typedef struct debug_vertex
+{
+	vec2 pos;
+	vec2 uv;
+	vec4 cubic;
+	vec4 color;
+	vec4 clip;
+	int zIndex;
+	u8 align2[12];
+} debug_vertex;
+
+
+#define LayoutNext(prevName, prevType, nextType) \
+	AlignUpOnPow2(_cat3_(LAYOUT_, prevName, _OFFSET)+_cat3_(LAYOUT_, prevType, _SIZE), _cat3_(LAYOUT_, nextType, _ALIGN))
+
+enum {
+	LAYOUT_VEC2_SIZE = 8,
+	LAYOUT_VEC2_ALIGN = 8,
+	LAYOUT_VEC4_SIZE = 16,
+	LAYOUT_VEC4_ALIGN = 16,
+	LAYOUT_INT_SIZE = 4,
+	LAYOUT_INT_ALIGN = 4,
+
+	LAYOUT_POS_OFFSET = 0,
+	LAYOUT_UV_OFFSET = LayoutNext(POS, VEC2, VEC2),
+	LAYOUT_CUBIC_OFFSET = LayoutNext(UV, VEC2, VEC4),
+	LAYOUT_COLOR_OFFSET = LayoutNext(CUBIC, VEC4, VEC4),
+	LAYOUT_CLIP_OFFSET = LayoutNext(COLOR, VEC4, VEC4),
+	LAYOUT_ZINDEX_OFFSET = LayoutNext(CLIP, VEC4, INT),
+
+	LAYOUT_VERTEX_ALIGN = 16,
+	LAYOUT_VERTEX_SIZE = LayoutNext(ZINDEX, INT, VERTEX),
+};
+
+enum {
+	MG_GL_CANVAS_DEFAULT_BUFFER_LENGTH = 1<<20,
+	MG_GL_CANVAS_VERTEX_BUFFER_SIZE = MG_GL_CANVAS_DEFAULT_BUFFER_LENGTH * LAYOUT_VERTEX_SIZE,
+	MG_GL_CANVAS_INDEX_BUFFER_SIZE = MG_GL_CANVAS_DEFAULT_BUFFER_LENGTH * LAYOUT_INT_SIZE,
+	MG_GL_CANVAS_TILE_COUNTER_BUFFER_SIZE = 65536,
+	MG_GL_CANVAS_TILE_ARRAY_SIZE = sizeof(int)*4096,
+	MG_GL_CANVAS_TILE_ARRAY_BUFFER_SIZE = MG_GL_CANVAS_TILE_COUNTER_BUFFER_SIZE * MG_GL_CANVAS_TILE_ARRAY_SIZE,
+};
+
+void mg_gl_canvas_update_vertex_layout(mg_gl_canvas_backend* backend)
+{
+	backend->interface.vertexLayout = (mg_vertex_layout){
+		    .maxVertexCount = MG_GL_CANVAS_DEFAULT_BUFFER_LENGTH,
+	        .maxIndexCount = MG_GL_CANVAS_DEFAULT_BUFFER_LENGTH,
+	        .posBuffer = backend->vertexMapping + LAYOUT_POS_OFFSET,
+	        .posStride = LAYOUT_VERTEX_SIZE,
+	        .cubicBuffer = backend->vertexMapping + LAYOUT_CUBIC_OFFSET,
+	        .cubicStride = LAYOUT_VERTEX_SIZE,
+	        .uvBuffer = backend->vertexMapping + LAYOUT_UV_OFFSET,
+	        .uvStride = LAYOUT_VERTEX_SIZE,
+	        .colorBuffer = backend->vertexMapping + LAYOUT_COLOR_OFFSET,
+	        .colorStride = LAYOUT_VERTEX_SIZE,
+	        .clipBuffer = backend->vertexMapping + LAYOUT_CLIP_OFFSET,
+	        .clipStride = LAYOUT_VERTEX_SIZE,
+	        .zIndexBuffer = backend->vertexMapping + LAYOUT_ZINDEX_OFFSET,
+	        .zIndexStride = LAYOUT_VERTEX_SIZE,
+	        .indexBuffer = backend->indexMapping,
+	        .indexStride = LAYOUT_INT_SIZE};
+}
+
+void mg_gl_send_buffers(mg_gl_canvas_backend* backend, int vertexCount, int indexCount)
+{
+	glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->vertexBuffer);
+	glBufferData(GL_SHADER_STORAGE_BUFFER, LAYOUT_VERTEX_SIZE*vertexCount, backend->vertexMapping, GL_DYNAMIC_DRAW);
+
+	glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->indexBuffer);
+	glBufferData(GL_SHADER_STORAGE_BUFFER, LAYOUT_INT_SIZE*indexCount, backend->indexMapping, GL_DYNAMIC_DRAW);
+}
+void mg_gl_canvas_begin(mg_canvas_backend* interface)
+{
+	mg_gl_canvas_backend* backend = (mg_gl_canvas_backend*)interface;
+	mg_gl_surface* surface = mg_gl_canvas_get_surface(backend);
+	if(!surface)
+	{
+		return;
+	}
+	glEnable(GL_BLEND);
+	glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
+}
+
+void mg_gl_canvas_end(mg_canvas_backend* interface)
+{
+	//NOTE: nothing to do here...
+}
+
+void mg_gl_canvas_clear(mg_canvas_backend* interface, mg_color clearColor)
+{
+	mg_gl_canvas_backend* backend = (mg_gl_canvas_backend*)interface;
+	mg_gl_surface* surface = mg_gl_canvas_get_surface(backend);
+	if(!surface)
+	{
+		return;
+	}
+
+	glClearColor(clearColor.r, clearColor.g, clearColor.b, clearColor.a);
+	glClear(GL_COLOR_BUFFER_BIT);
+}
+
+void mg_gl_canvas_draw_batch(mg_canvas_backend* interface, u32 vertexCount, u32 indexCount)
+{
+	mg_gl_canvas_backend* backend = (mg_gl_canvas_backend*)interface;
+	mg_gl_surface* surface = mg_gl_canvas_get_surface(backend);
+	if(!surface)
+	{
+		return;
+	}
+
+/*NOTE: if we want debug_vertex while debugging, the following ensures the struct def doesn't get stripped away
+	debug_vertex vertex;
+	printf("foo %p\n", &vertex);
+//*/
+	mg_gl_send_buffers(backend, vertexCount, indexCount);
+
+	mp_rect frame = mg_surface_get_frame(backend->surface);
+
+	const int tileSize = 16;
+	const int tileCountX = (frame.w + tileSize - 1)/tileSize;
+	const int tileCountY = (frame.h + tileSize - 1)/tileSize;
+	const int tileArraySize = MG_GL_CANVAS_TILE_ARRAY_SIZE;
+
+	//TODO: ensure there's enough space in tile buffer
+
+	//NOTE: first clear counters
+	glUseProgram(backend->clearCounterProgram);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->tileCounterBuffer);
+	glDispatchCompute(tileCountX*tileCountY, 1, 1);
+
+	//NOTE: we first distribute triangles into tiles:
+
+	glUseProgram(backend->tileProgram);
+
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->vertexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->indexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileCounterBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileArrayBuffer);
+
+	glUniform1ui(0, indexCount);
+	glUniform2ui(1, tileCountX, tileCountY);
+	glUniform1ui(2, tileSize);
+	glUniform1ui(3, tileArraySize);
+
+	u32 threadCount = indexCount/3;
+	glDispatchCompute((threadCount + 255)/256, 1, 1);
+
+	//NOTE: next we sort triangles in each tile
+	glUseProgram(backend->sortProgram);
+/*
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->vertexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->indexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileCounterBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileArrayBuffer);
+*/
+	glUniform1ui(0, indexCount);
+	glUniform2ui(1, tileCountX, tileCountY);
+	glUniform1ui(2, tileSize);
+	glUniform1ui(3, tileArraySize);
+
+	glDispatchCompute(tileCountX * tileCountY, 1, 1);
+
+	//TODO: then we fire the fragment shader that will select only triangles in its tile
+//	glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
+
+	glUseProgram(backend->drawProgram);
+/*
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->vertexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->indexBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileCounterBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileArrayBuffer);
+*/
+	glBindImageTexture(0, backend->outTexture, 0, GL_FALSE, 0, GL_WRITE_ONLY, GL_RGBA8);
+
+	glUniform1ui(0, indexCount);
+	glUniform2ui(1, tileCountX, tileCountY);
+	glUniform1ui(2, tileSize);
+	glUniform1ui(3, tileArraySize);
+
+	glDispatchCompute(tileCountX, tileCountY, 1);
+
+	//NOTE: now blit out texture to surface
+	glUseProgram(backend->blitProgram);
+	glBindBuffer(GL_ARRAY_BUFFER, backend->dummyVertexBuffer);
+	glActiveTexture(GL_TEXTURE0);
+	glBindTexture(GL_TEXTURE_2D, backend->outTexture);
+	glUniform1i(0, 0);
+
+	glDrawArrays(GL_TRIANGLES, 0, 6);
+
+	mg_gl_canvas_update_vertex_layout(backend);
+}
+
+void mg_gl_canvas_destroy(mg_canvas_backend* interface)
+{
+	mg_gl_canvas_backend* backend = (mg_gl_canvas_backend*)interface;
+
+	//TODO
+}
+
+void mg_gl_canvas_atlas_upload(mg_canvas_backend* interface, mp_rect rect, u8* bytes)
+{
+	//TODO
+}
+
+static void mg_gl_compile_shader(GLuint shader, const char* source)
+{
+	glShaderSource(shader, 1, &source, 0);
+	glCompileShader(shader);
+
+	int err = glGetError();
+	if(err)
+	{
+		printf("gl error: %i\n", err);
+	}
+
+	int status = 0;
+	glGetShaderiv(shader, GL_COMPILE_STATUS, &status);
+	if(!status)
+	{
+		char buffer[256];
+		int size = 0;
+		glGetShaderInfoLog(shader, 256, &size, buffer);
+		printf("shader error: %.*s\n", size, buffer);
+	}
+}
+
+mg_canvas_backend* mg_gl_canvas_create(mg_surface surface)
+{
+	mg_gl_canvas_backend* backend = 0;
+	mg_surface_data* surfaceData = mg_surface_data_from_handle(surface);
+	if(surfaceData && surfaceData->backend == MG_BACKEND_GL)
+	{
+		mg_gl_surface* glSurface = (mg_gl_surface*)surfaceData;
+
+		backend = malloc_type(mg_gl_canvas_backend);
+		memset(backend, 0, sizeof(mg_gl_canvas_backend));
+		backend->surface = surface;
+
+		//NOTE(martin): setup interface functions
+		backend->interface.destroy = mg_gl_canvas_destroy;
+		backend->interface.begin = mg_gl_canvas_begin;
+		backend->interface.end = mg_gl_canvas_end;
+		backend->interface.clear = mg_gl_canvas_clear;
+		backend->interface.drawBatch = mg_gl_canvas_draw_batch;
+		backend->interface.atlasUpload = mg_gl_canvas_atlas_upload;
+
+		mg_surface_prepare(surface);
+
+		GLuint vao;
+		glGenVertexArrays(1, &vao);
+		glBindVertexArray(vao);
+
+		glGenBuffers(1, &backend->dummyVertexBuffer);
+		glBindBuffer(GL_ARRAY_BUFFER, backend->dummyVertexBuffer);
+
+		glGenBuffers(1, &backend->vertexBuffer);
+		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->vertexBuffer);
+//		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_CANVAS_VERTEX_BUFFER_SIZE, 0, GL_DYNAMIC_DRAW);
+
+		glGenBuffers(1, &backend->indexBuffer);
+		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->indexBuffer);
+//		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_CANVAS_INDEX_BUFFER_SIZE, 0, GL_DYNAMIC_DRAW);
+
+		glGenBuffers(1, &backend->tileCounterBuffer);
+		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->tileCounterBuffer);
+		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_CANVAS_TILE_COUNTER_BUFFER_SIZE, 0, GL_DYNAMIC_COPY);
+
+		glGenBuffers(1, &backend->tileArrayBuffer);
+		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->tileArrayBuffer);
+		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_CANVAS_TILE_ARRAY_BUFFER_SIZE, 0, GL_DYNAMIC_COPY);
+
+		mp_rect frame = mg_surface_get_frame(backend->surface);
+		glGenTextures(1, &backend->outTexture);
+		glBindTexture(GL_TEXTURE_2D, backend->outTexture);
+		glTexStorage2D(GL_TEXTURE_2D, 1, GL_RGBA8, frame.w, frame.h);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+
+		//NOTE: create clear program
+		{
+			GLuint clearShader = glCreateShader(GL_COMPUTE_SHADER);
+			backend->clearCounterProgram = glCreateProgram();
+
+			mg_gl_compile_shader(clearShader, gles_canvas_clear_counters);
+
+			glAttachShader(backend->clearCounterProgram, clearShader);
+			glLinkProgram(backend->clearCounterProgram);
+
+			int status = 0;
+			glGetProgramiv(backend->clearCounterProgram, GL_LINK_STATUS, &status);
+			if(!status)
+			{
+				char buffer[256];
+				int size = 0;
+				glGetProgramInfoLog(backend->clearCounterProgram, 256, &size, buffer);
+				printf("link error in gl_canvas_clear_counters: %.*s\n", size, buffer);
+				exit(-1);
+ 			}
+		}
+		//NOTE: create tile program
+		{
+			GLuint tileShader = glCreateShader(GL_COMPUTE_SHADER);
+			backend->tileProgram = glCreateProgram();
+
+			mg_gl_compile_shader(tileShader, gles_canvas_tile);
+
+			glAttachShader(backend->tileProgram, tileShader);
+			glLinkProgram(backend->tileProgram);
+
+			int status = 0;
+			glGetProgramiv(backend->tileProgram, GL_LINK_STATUS, &status);
+			if(!status)
+			{
+				char buffer[256];
+				int size = 0;
+				glGetProgramInfoLog(backend->tileProgram, 256, &size, buffer);
+				printf("link error in gl_canvas_tile: %.*s\n", size, buffer);
+				exit(-1);
+ 			}
+		}
+		//NOTE: create sort program
+		{
+			GLuint sortShader = glCreateShader(GL_COMPUTE_SHADER);
+			backend->sortProgram = glCreateProgram();
+
+			mg_gl_compile_shader(sortShader, gles_canvas_sort);
+
+			glAttachShader(backend->sortProgram, sortShader);
+			glLinkProgram(backend->sortProgram);
+
+			int status = 0;
+			glGetProgramiv(backend->sortProgram, GL_LINK_STATUS, &status);
+			if(!status)
+			{
+				char buffer[256];
+				int size = 0;
+				glGetProgramInfoLog(backend->sortProgram, 256, &size, buffer);
+				printf("link error gl_canvas_sort: %.*s\n", size, buffer);
+				exit(-1);
+ 			}
+		}
+
+		//NOTE: create draw program
+		{
+			GLuint shader = glCreateShader(GL_COMPUTE_SHADER);
+			backend->drawProgram = glCreateProgram();
+
+			mg_gl_compile_shader(shader, gles_canvas_draw);
+
+			glAttachShader(backend->drawProgram, shader);
+			glLinkProgram(backend->drawProgram);
+
+			int status = 0;
+			glGetProgramiv(backend->drawProgram, GL_LINK_STATUS, &status);
+			if(!status)
+			{
+				char buffer[256];
+				int size = 0;
+				glGetProgramInfoLog(backend->drawProgram, 256, &size, buffer);
+				printf("link error gl_canvas_draw: %.*s\n", size, buffer);
+				exit(-1);
+ 			}
+		}
+
+		//NOTE: create blit program
+		{
+			GLuint vertexShader = glCreateShader(GL_VERTEX_SHADER);
+			GLuint fragmentShader = glCreateShader(GL_FRAGMENT_SHADER);
+			backend->blitProgram = glCreateProgram();
+
+			mg_gl_compile_shader(vertexShader, gles_canvas_blit_vertex);
+			mg_gl_compile_shader(fragmentShader, gles_canvas_blit_fragment);
+
+			glAttachShader(backend->blitProgram, vertexShader);
+			glAttachShader(backend->blitProgram, fragmentShader);
+			glLinkProgram(backend->blitProgram);
+
+			int status = 0;
+			glGetProgramiv(backend->blitProgram, GL_LINK_STATUS, &status);
+			if(!status)
+			{
+				char buffer[256];
+				int size = 0;
+				glGetProgramInfoLog(backend->blitProgram, 256, &size, buffer);
+				printf("link error gl_canvas_blit: %.*s\n", size, buffer);
+				exit(-1);
+ 			}
+		}
+
+		backend->vertexMapping = malloc_array(char, 1<<30);
+		backend->indexMapping = malloc_array(char, 1<<30);
+
+		mg_gl_canvas_update_vertex_layout(backend);
+	}
+
+	return((mg_canvas_backend*)backend);
+}
+
+
+#undef LOG_SUBSYSTEM
diff --git a/src/gles_canvas_shaders/gles_canvas_blit_fragment.glsl b/src/gles_canvas_shaders/gles_canvas_blit_fragment.glsl
index 910bd96..4a51e91 100644
--- a/src/gles_canvas_shaders/gles_canvas_blit_fragment.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_blit_fragment.glsl
@@ -1,4 +1,4 @@
-#version 310 es
+#version 430
 
 precision mediump float;
 
diff --git a/src/gles_canvas_shaders/gles_canvas_blit_vertex.glsl b/src/gles_canvas_shaders/gles_canvas_blit_vertex.glsl
index e4af48a..5547012 100644
--- a/src/gles_canvas_shaders/gles_canvas_blit_vertex.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_blit_vertex.glsl
@@ -1,4 +1,4 @@
-#version 310 es
+#version 430
 
 precision mediump float;
 
diff --git a/src/gles_canvas_shaders/gles_canvas_clear_counters.glsl b/src/gles_canvas_shaders/gles_canvas_clear_counters.glsl
index 9e7fd01..321bf38 100644
--- a/src/gles_canvas_shaders/gles_canvas_clear_counters.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_clear_counters.glsl
@@ -1,4 +1,4 @@
-#version 310 es
+#version 430
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 precision mediump float;
diff --git a/src/gles_canvas_shaders/gles_canvas_draw.glsl b/src/gles_canvas_shaders/gles_canvas_draw.glsl
index 49a0512..fa0f0ae 100644
--- a/src/gles_canvas_shaders/gles_canvas_draw.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_draw.glsl
@@ -1,8 +1,8 @@
-#version 310 es
+#version 430
 layout(local_size_x = 16, local_size_y = 16, local_size_z = 1) in;
 
 precision mediump float;
-precision mediump image2D;
+//precision mediump image2D;
 
 layout(std430) buffer;
 
@@ -59,7 +59,7 @@ void main()
 	const float subPixelFactor = 16.;
 	ivec2 centerPoint = ivec2(round((vec2(pixelCoord) + vec2(0.5, 0.5)) * subPixelFactor));
 
-/*
+//*
 	const int sampleCount = 8;
 	ivec2 samplePoints[sampleCount] = ivec2[sampleCount](centerPoint + ivec2(1, 3),
 	                                                     centerPoint + ivec2(-1, -3),
@@ -69,12 +69,13 @@ void main()
 	                                                     centerPoint + ivec2(-7, 1),
 	                                                     centerPoint + ivec2(3, -7),
 	                                                     centerPoint + ivec2(7, 7));
-*/
+/*/
 	const int sampleCount = 4;
 	ivec2 samplePoints[sampleCount] = ivec2[sampleCount](centerPoint + ivec2(-2, 6),
 	                                                     centerPoint + ivec2(6, 2),
 	                                                     centerPoint + ivec2(-6, -2),
 	                                                     centerPoint + ivec2(2, -6));
+//*/
 	//DEBUG
 /*
 	{
diff --git a/src/gles_canvas_shaders/gles_canvas_sort.glsl b/src/gles_canvas_shaders/gles_canvas_sort.glsl
index d7648f9..4b58e28 100644
--- a/src/gles_canvas_shaders/gles_canvas_sort.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_sort.glsl
@@ -1,4 +1,4 @@
-#version 310 es
+#version 430
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 precision mediump float;
diff --git a/src/gles_canvas_shaders/gles_canvas_tile.glsl b/src/gles_canvas_shaders/gles_canvas_tile.glsl
index 5d2d577..a96fe76 100644
--- a/src/gles_canvas_shaders/gles_canvas_tile.glsl
+++ b/src/gles_canvas_shaders/gles_canvas_tile.glsl
@@ -1,4 +1,4 @@
-#version 310 es
+#version 430
 layout(local_size_x = 512, local_size_y = 1, local_size_z = 1) in;
 
 precision mediump float;
diff --git a/src/graphics.c b/src/graphics.c
index 3747985..d65e094 100644
--- a/src/graphics.c
+++ b/src/graphics.c
@@ -2610,6 +2610,11 @@ mp_rect mg_text_bounding_box(mg_font font, f32 fontSize, str8 text)
 	mg_canvas_backend* mg_gles_canvas_create(mg_surface surface);
 #endif
 
+#ifdef MG_IMPLEMENTS_BACKEND_GL
+	mg_canvas_backend* mg_gl_canvas_create(mg_surface surface);
+#endif
+
+
 mg_canvas mg_canvas_create(mg_surface surface)
 {
 	mg_canvas canvas = mg_canvas_nil();
@@ -2631,11 +2636,9 @@ mg_canvas mg_canvas_create(mg_surface surface)
 				break;
 		#endif
 
-		/*
-			case MG_BACKEND_OPENGL:
-				canvasData = mg_opengl_canvas_create(surface);
+			case MG_BACKEND_GL:
+				backend = mg_gl_canvas_create(surface);
 				break;
-		*/
 			default:
 				break;
 		}
diff --git a/src/milepost.c b/src/milepost.c
index 5a6b16c..e252a89 100644
--- a/src/milepost.c
+++ b/src/milepost.c
@@ -53,9 +53,10 @@
 
 #if defined(OS_WIN64)
 	#include"win32_app.c"
-//	#include"win32_gl_surface.c"
-	#include"win32_gles_surface.c"
-	#include"gles_canvas.c"
+	#include"win32_gl_surface.c"
+	#include"gl_canvas.c"
+//	#include"win32_gles_surface.c"
+//	#include"gles_canvas.c"
 #elif defined(OS_MACOS)
 	//NOTE: macos application layer is defined in milepost.m
 #else
diff --git a/src/milepost.h b/src/milepost.h
index ed87c6a..e3f5ffc 100644
--- a/src/milepost.h
+++ b/src/milepost.h
@@ -38,12 +38,18 @@
 #include"mp_app.h"
 
 #if defined(OS_WIN64) || defined(OS_WIN32)
-	#define WIN32_GL_LOADER_API
-//	#include"win32_gl_loader.h"
 
 	#if MG_IMPLEMENTS_BACKEND_GLES
 		#include"win32_gles_surface.h"
 	#endif
+
+	#if MG_IMPLEMENTS_BACKEND_GL
+		#define WIN32_GL_LOADER_API
+		#include"win32_gl_loader.h"
+		#undef WIN32_GL_LOADER_API
+
+		#include"win32_gl_surface.h"
+	#endif
 #endif
 
 #if MG_IMPLEMENTS_BACKEND_METAL
diff --git a/src/win32_gl_loader.h b/src/win32_gl_loader.h
index 24f6ad0..769a924 100644
--- a/src/win32_gl_loader.h
+++ b/src/win32_gl_loader.h
@@ -1,65 +1,73 @@
-/************************************************************//**
-*
-*	@file: win32_gl_loader.h
-*	@author: Martin Fouilleul
-*	@date: 01/08/2022
-*	@revision:
-*
-*****************************************************************/
-
-#define WIN32_LEAN_AND_MEAN
-#include<windows.h>
-#include<GL/gl.h>
-#include"GL/glext.h"
-#include"GL/wglext.h"
-
-#include"macro_helpers.h"
-
-#define GL_PROC_LIST \
-	GL_PROC(WGLCHOOSEPIXELFORMATARB, wglChoosePixelFormatARB) \
-	GL_PROC(WGLCREATECONTEXTATTRIBSARB, wglCreateContextAttribsARB) \
-	GL_PROC(WGLMAKECONTEXTCURRENTARB, wglMakeContextCurrentARB) \
-	GL_PROC(WGLSWAPINTERVALEXT, wglSwapIntervalEXT) \
-	GL_PROC(GLCREATESHADER, glCreateShader) \
-	GL_PROC(GLCREATEPROGRAM, glCreateProgram) \
-	GL_PROC(GLATTACHSHADER, glAttachShader) \
-	GL_PROC(GLCOMPILESHADER, glCompileShader) \
-	GL_PROC(GLGETSHADERIV, glGetShaderiv) \
-	GL_PROC(GLGETSHADERINFOLOG, glGetShaderInfoLog) \
-	GL_PROC(GLSHADERSOURCE, glShaderSource) \
-	GL_PROC(GLLINKPROGRAM, glLinkProgram) \
-	GL_PROC(GLGETPROGRAMIV, glGetProgramiv) \
-	GL_PROC(GLGETPROGRAMINFOLOG, glGetProgramInfoLog) \
-	GL_PROC(GLUSEPROGRAM, glUseProgram) \
-	GL_PROC(GLGENVERTEXARRAYS, glGenVertexArrays) \
-	GL_PROC(GLBINDVERTEXARRAY, glBindVertexArray) \
-	GL_PROC(GLGENBUFFERS, glGenBuffers) \
-	GL_PROC(GLBINDBUFFER, glBindBuffer) \
-	GL_PROC(GLBUFFERDATA, glBufferData) \
-	GL_PROC(GLUNIFORMMATRIX4FV, glUniformMatrix4fv) \
-	GL_PROC(GLVERTEXATTRIBPOINTER, glVertexAttribPointer) \
-	GL_PROC(GLENABLEVERTEXATTRIBARRAY, glEnableVertexAttribArray)
-
-#ifdef WIN32_GL_LOADER_API
-	//NOTE: pointer declarations
-	#define GL_PROC(type, name) extern _cat3_(PFN, type, PROC) name;
-	GL_PROC_LIST
-	#undef GL_PROC
-#endif
-
-
-#ifdef WIN32_GL_LOADER_IMPL
-#define GL_PROC(type, name) _cat3_(PFN, type, PROC) name = 0;
-	GL_PROC_LIST
-#undef GL_PROC
-
-void mp_gl_load_procs()
-{
-	#define GL_PROC(type, name) name = (_cat3_(PFN, type, PROC))wglGetProcAddress( #name );
-		GL_PROC_LIST
-	#undef GL_PROC
-}
-
-#endif
-
-#undef GL_PROC_LIST
+/************************************************************//**
+*
+*	@file: win32_gl_loader.h
+*	@author: Martin Fouilleul
+*	@date: 01/08/2022
+*	@revision:
+*
+*****************************************************************/
+
+#define WIN32_LEAN_AND_MEAN
+#include<windows.h>
+#include<GL/gl.h>
+#include<GL/glext.h>
+#include<GL/wglext.h>
+
+#include"macro_helpers.h"
+
+#define GL_PROC_LIST \
+	GL_PROC(WGLCHOOSEPIXELFORMATARB, wglChoosePixelFormatARB) \
+	GL_PROC(WGLCREATECONTEXTATTRIBSARB, wglCreateContextAttribsARB) \
+	GL_PROC(WGLMAKECONTEXTCURRENTARB, wglMakeContextCurrentARB) \
+	GL_PROC(WGLSWAPINTERVALEXT, wglSwapIntervalEXT) \
+	GL_PROC(GLCREATESHADER, glCreateShader) \
+	GL_PROC(GLCREATEPROGRAM, glCreateProgram) \
+	GL_PROC(GLATTACHSHADER, glAttachShader) \
+	GL_PROC(GLCOMPILESHADER, glCompileShader) \
+	GL_PROC(GLGETSHADERIV, glGetShaderiv) \
+	GL_PROC(GLGETSHADERINFOLOG, glGetShaderInfoLog) \
+	GL_PROC(GLSHADERSOURCE, glShaderSource) \
+	GL_PROC(GLLINKPROGRAM, glLinkProgram) \
+	GL_PROC(GLGETPROGRAMIV, glGetProgramiv) \
+	GL_PROC(GLGETPROGRAMINFOLOG, glGetProgramInfoLog) \
+	GL_PROC(GLUSEPROGRAM, glUseProgram) \
+	GL_PROC(GLGENVERTEXARRAYS, glGenVertexArrays) \
+	GL_PROC(GLBINDVERTEXARRAY, glBindVertexArray) \
+	GL_PROC(GLGENBUFFERS, glGenBuffers) \
+	GL_PROC(GLBINDBUFFER, glBindBuffer) \
+	GL_PROC(GLBUFFERDATA, glBufferData) \
+	GL_PROC(GLUNIFORMMATRIX4FV, glUniformMatrix4fv) \
+	GL_PROC(GLVERTEXATTRIBPOINTER, glVertexAttribPointer) \
+	GL_PROC(GLENABLEVERTEXATTRIBARRAY, glEnableVertexAttribArray) \
+	GL_PROC(GLBINDBUFFERBASE, glBindBufferBase) \
+	GL_PROC(GLDISPATCHCOMPUTE, glDispatchCompute) \
+	GL_PROC(GLUNIFORM1UI, glUniform1ui) \
+	GL_PROC(GLUNIFORM2UI, glUniform2ui) \
+	GL_PROC(GLBINDIMAGETEXTURE, glBindImageTexture) \
+	GL_PROC(GLACTIVETEXTURE, glActiveTexture) \
+	GL_PROC(GLUNIFORM1I, glUniform1i) \
+	GL_PROC(GLTEXSTORAGE2D, glTexStorage2D) \
+
+#ifdef WIN32_GL_LOADER_API
+	//NOTE: pointer declarations
+	#define GL_PROC(type, name) extern _cat3_(PFN, type, PROC) name;
+	GL_PROC_LIST
+	#undef GL_PROC
+#endif
+
+
+#ifdef WIN32_GL_LOADER_IMPL
+#define GL_PROC(type, name) _cat3_(PFN, type, PROC) name = 0;
+	GL_PROC_LIST
+#undef GL_PROC
+
+void mp_gl_load_procs()
+{
+	#define GL_PROC(type, name) name = (_cat3_(PFN, type, PROC))wglGetProcAddress( #name );
+		GL_PROC_LIST
+	#undef GL_PROC
+}
+
+#endif
+
+#undef GL_PROC_LIST
diff --git a/src/win32_gl_surface.c b/src/win32_gl_surface.c
index 1b31e70..f732cdc 100644
--- a/src/win32_gl_surface.c
+++ b/src/win32_gl_surface.c
@@ -13,12 +13,11 @@
 #include"graphics_internal.h"
 #include"win32_app.h"
 
-#define MG_IMPLEMENTS_BACKEND_GL
-
 typedef struct mg_gl_surface
 {
 	mg_surface_data interface;
 
+	HWND hWnd;
 	HDC hDC;
 	HGLRC glContext;
 
@@ -44,30 +43,15 @@ void mg_gl_surface_present(mg_surface_data* interface)
 	SwapBuffers(surface->hDC);
 }
 
-PFNWGLCHOOSEPIXELFORMATARBPROC wglChoosePixelFormatARB;
-PFNWGLCREATECONTEXTATTRIBSARBPROC wglCreateContextAttribsARB;
-PFNWGLMAKECONTEXTCURRENTARBPROC wglMakeContextCurrentARB;
-PFNWGLSWAPINTERVALEXTPROC wglSwapIntervalEXT;
+mp_rect mg_gl_surface_get_frame(mg_surface_data* interface)
+{
+	mg_gl_surface* surface = (mg_gl_surface*)interface;
+	RECT rect = {0};
+	GetClientRect(surface->hWnd, &rect);
 
-PFNGLCREATESHADERPROC glCreateShader;
-PFNGLCREATEPROGRAMPROC glCreateProgram;
-PFNGLATTACHSHADERPROC glAttachShader;
-PFNGLCOMPILESHADERPROC glCompileShader;
-PFNGLGETSHADERIVPROC glGetShaderiv;
-PFNGLGETSHADERINFOLOGPROC glGetShaderInfoLog;
-PFNGLSHADERSOURCEPROC glShaderSource;
-PFNGLLINKPROGRAMPROC glLinkProgram;
-PFNGLGETPROGRAMIVPROC glGetProgramiv;
-PFNGLGETPROGRAMINFOLOGPROC glGetProgramInfoLog;
-PFNGLUSEPROGRAMPROC glUseProgram;
-PFNGLGENVERTEXARRAYSPROC glGenVertexArrays;
-PFNGLBINDVERTEXARRAYPROC glBindVertexArray;
-PFNGLGENBUFFERSPROC glGenBuffers;
-PFNGLBINDBUFFERPROC glBindBuffer;
-PFNGLBUFFERDATAPROC glBufferData;
-PFNGLUNIFORMMATRIX4FVPROC glUniformMatrix4fv;
-PFNGLVERTEXATTRIBPOINTERPROC glVertexAttribPointer;
-PFNGLENABLEVERTEXATTRIBARRAYPROC glEnableVertexAttribArray;
+	mp_rect res = {rect.left, rect.bottom, rect.right - rect.left, rect.bottom - rect.top};
+	return(res);
+}
 
 mg_surface mg_gl_surface_create_for_window(mp_window window)
 {
@@ -130,6 +114,7 @@ mg_surface mg_gl_surface_create_for_window(mp_window window)
 		wglMakeCurrent(dummyDC, dummyGLContext);
 
 		//NOTE(martin): now load extension functions
+		/*
 		wglChoosePixelFormatARB = (PFNWGLCHOOSEPIXELFORMATARBPROC)wglGetProcAddress("wglChoosePixelFormatARB");
 		wglCreateContextAttribsARB = (PFNWGLCREATECONTEXTATTRIBSARBPROC)wglGetProcAddress("wglCreateContextAttribsARB");
 		wglMakeContextCurrentARB = (PFNWGLMAKECONTEXTCURRENTARBPROC)wglGetProcAddress("wglMakeContextCurrentARB");
@@ -154,6 +139,8 @@ mg_surface mg_gl_surface_create_for_window(mp_window window)
 		glUniformMatrix4fv = (PFNGLUNIFORMMATRIX4FVPROC)wglGetProcAddress("glUniformMatrix4fv");
 		glVertexAttribPointer = (PFNGLVERTEXATTRIBPOINTERPROC)wglGetProcAddress("glVertexAttribPointer");
 		glEnableVertexAttribArray = (PFNGLENABLEVERTEXATTRIBARRAYPROC)wglGetProcAddress("glEnableVertexAttribArray");
+		*/
+		mp_gl_load_procs();
 
 		//NOTE(martin): now create the true pixel format and gl context
 		int pixelFormatAttrs[] = {
@@ -180,6 +167,7 @@ mg_surface mg_gl_surface_create_for_window(mp_window window)
 		int contextAttrs[] = {
 			WGL_CONTEXT_MAJOR_VERSION_ARB, 4,
 			WGL_CONTEXT_MINOR_VERSION_ARB, 3,
+			WGL_CONTEXT_PROFILE_MASK_ARB, WGL_CONTEXT_CORE_PROFILE_BIT_ARB,
 			0};
 
 		HGLRC glContext = wglCreateContextAttribsARB(hDC, dummyGLContext, contextAttrs);
@@ -198,7 +186,9 @@ mg_surface mg_gl_surface_create_for_window(mp_window window)
 
 		//NOTE: make gl context current
 		wglMakeCurrent(hDC, glContext);
-		wglSwapIntervalEXT(1);
+
+		//TODO: set this back to 1 after testing
+		wglSwapIntervalEXT(0);
 
 		//TODO save important info in surface_data and return a handle
 		mg_gl_surface* surface = malloc_type(mg_gl_surface);
@@ -206,9 +196,10 @@ mg_surface mg_gl_surface_create_for_window(mp_window window)
 		surface->interface.destroy = mg_gl_surface_destroy;
 		surface->interface.prepare = mg_gl_surface_prepare;
 		surface->interface.present = mg_gl_surface_present;
+		surface->interface.getFrame = mg_gl_surface_get_frame;
 
 		//TODO: get/set frame/hidden
-
+		surface->hWnd = windowData->win32.hWnd;
 		surface->hDC = hDC;
 		surface->glContext = glContext;
 
diff --git a/src/win32_gl_surface.h b/src/win32_gl_surface.h
index a212b03..1e130ee 100644
--- a/src/win32_gl_surface.h
+++ b/src/win32_gl_surface.h
@@ -9,6 +9,8 @@
 #ifndef __WIN32_GL_SURFACE_H_
 #define __WIN32_GL_SURFACE_H_
 
+#include"graphics.h"
+
 mg_surface mg_gl_surface_create_for_window(mp_window window);
 
 #endif // __WIN32_GL_SURFACE_H_
diff --git a/vector_renderer_notes.txt b/vector_renderer_notes.txt
index b884241..d3a79e5 100644
--- a/vector_renderer_notes.txt
+++ b/vector_renderer_notes.txt
@@ -68,3 +68,15 @@ Notes:
 	* Doing glBufferData from a small build buffer is surprisingly slow...
 		* Angle seems to take into account only the first call to glBufferData to allocate size, and then send the full buffer???
 		* Not pre-allocating in creation procedure "solves" the problem with glBufferData...
+
+Only way seems to be to reduce buffer size:
+	* pack vec2 together to avoid padding
+	* store color, clip and uv out of band?
+		* eg have a shape buffer containing uv for bounding box, clip, and color?
+		* could also still store color as an int "for free" next to shape index
+		* could send pos and uv as half-floats?
+
+Ideally, we would do the vertex computation on the GPU (opportunity to parallelize AND avoid sending vertex buffer...)
+
+-> Problem profiling canvas renderer with angle (not really well supported in renderdoc)
+	-> Maybe use an OpenGL backend instead?