hello_triangle_verbose.c

#include <assert.h>
#include <stdio.h>
#include <memory.h>
#include <miniprintf.h>
#include "lib_webgpu.h"

WGpuAdapter adapter;
WGpuCanvasContext canvasContext;
WGpuDevice device;
WGpuQueue queue;
WGpuRenderPipeline renderPipeline;

void uncapturedError(WGpuDevice device, WGPU_ERROR_TYPE errorType, const char *errorMessage, void *userData)
{
  emscripten_mini_stdio_fprintf(EM_STDERR, "Uncaptured WebGPU error: type: %d, message: %s\n", errorType, errorMessage);
}

void onSubmittedWorkDone(WGpuQueue queue, void *userData)
{
  emscripten_mini_stdio_printf("Submitted work done callback fired! (queue=%u, userData=%u)\n", queue, userData);
}

EM_BOOL raf(double time, void *userData)
{
  WGpuCommandEncoder encoder = wgpu_device_create_command_encoder(device, 0);
  assert(wgpu_is_command_encoder(encoder));

  WGpuTexture swapChainTexture = wgpu_canvas_context_get_current_texture(canvasContext);
  assert(wgpu_is_texture(swapChainTexture));

  // Calling .getCurrentTexture() several times within a single rAF() callback
  // should return the same binding to the swap chain texture.
  WGpuTexture swapChainTexture2 = wgpu_canvas_context_get_current_texture(canvasContext);
  assert(swapChainTexture == swapChainTexture2);
  (void)swapChainTexture2;

  WGpuRenderPassColorAttachment colorAttachment = WGPU_RENDER_PASS_COLOR_ATTACHMENT_DEFAULT_INITIALIZER;
  colorAttachment.view = wgpu_texture_create_view(swapChainTexture, 0);
  assert(wgpu_is_texture_view(colorAttachment.view));

  WGpuRenderPassDescriptor passDesc = {};
  passDesc.numColorAttachments = 1;
  passDesc.colorAttachments = &colorAttachment;

  WGpuRenderPassEncoder pass = wgpu_command_encoder_begin_render_pass(encoder, &passDesc);
  assert(wgpu_is_render_pass_encoder(pass));
  wgpu_render_pass_encoder_set_pipeline(pass, renderPipeline);
  wgpu_render_pass_encoder_draw(pass, 3, 1, 0, 0);
  wgpu_render_pass_encoder_end(pass);

  WGpuCommandBuffer commandBuffer = wgpu_command_encoder_finish(encoder);
  assert(wgpu_is_command_buffer(commandBuffer));

  wgpu_queue_submit_one_and_destroy(queue, commandBuffer);

  wgpu_queue_set_on_submitted_work_done_callback(queue, onSubmittedWorkDone, 0);

  static int numLiveObjects = 0;
  int numLiveNow = wgpu_get_num_live_objects();
  if (numLiveNow != numLiveObjects)
  {
    emscripten_mini_stdio_printf("Num live WebGPU objects: %u\n", numLiveNow);
    numLiveObjects = numLiveNow;
  }
  assert(wgpu_get_num_live_objects() < 100); // Check against programming errors from Wasm<->JS WebGPU object leaks

  return EM_FALSE;//EM_TRUE; // This is static content, but keep rendering to debug leaking WebGPU objects above
}

void ObtainedWebGpuDevice(WGpuDevice result, void *userData)
{
  assert(userData == (void*)43);
  assert(wgpu_is_device(result));
  device = result;

  wgpu_device_set_uncapturederror_callback(device, uncapturedError, 0);

  char deviceLabel[256];
  memset(deviceLabel, 0xEE, sizeof(deviceLabel));
  wgpu_object_get_label(device, deviceLabel, sizeof(deviceLabel));
  assert(strlen(deviceLabel) == 0); // Initial label should be empty.

  wgpu_object_set_label(device, "My WebGPU device");
  wgpu_object_get_label(device, deviceLabel, sizeof(deviceLabel));
  printf("Got device, set label: \"%s\"\n", deviceLabel);
  assert(!strcmp(deviceLabel, "My WebGPU device"));

  queue = wgpu_device_get_queue(device);
  assert(wgpu_is_queue(queue));

  // TODO: read device.features and device.limits;

  canvasContext = wgpu_canvas_get_webgpu_context("canvas");
  assert(canvasContext);
  assert(wgpu_is_canvas_context(canvasContext));

  WGpuCanvasConfiguration config = WGPU_CANVAS_CONFIGURATION_DEFAULT_INITIALIZER;
  config.device = device;
  config.format = navigator_gpu_get_preferred_canvas_format();
  // TODO
//  emscripten_mini_stdio_printf("Preferred swap chain format: %s\n", wgpu_enum_to_string(config.format));

  wgpu_canvas_context_configure(canvasContext, &config);

  const char *vertexShader =
    "@vertex\n"
    "fn main(@builtin(vertex_index) vertexIndex : u32) -> @builtin(position) vec4<f32> {\n"
      "var pos = array<vec2<f32>, 3>(\n"
        "vec2<f32>(0.0, 0.5),\n"
        "vec2<f32>(-0.5, -0.5),\n"
        "vec2<f32>(0.5, -0.5)\n"
      ");\n"

      "return vec4<f32>(pos[vertexIndex], 0.0, 1.0);\n"
    "}\n";

  const char *fragmentShader =
    "@fragment\n"
    "fn main() -> @location(0) vec4<f32> {\n"
      "return vec4<f32>(1.0, 0.5, 0.3, 1.0);\n"
    "}\n";

  WGpuShaderModuleDescriptor shaderModuleDesc = {};
  shaderModuleDesc.code = vertexShader;
  WGpuShaderModule vs = wgpu_device_create_shader_module(device, &shaderModuleDesc);
  assert(wgpu_is_shader_module(vs));

  shaderModuleDesc.code = fragmentShader;
  WGpuShaderModule fs = wgpu_device_create_shader_module(device, &shaderModuleDesc);
  assert(wgpu_is_shader_module(fs));

  WGpuRenderPipelineDescriptor renderPipelineDesc = WGPU_RENDER_PIPELINE_DESCRIPTOR_DEFAULT_INITIALIZER;
  renderPipelineDesc.vertex.module = vs;
  renderPipelineDesc.vertex.entryPoint = "main";
  renderPipelineDesc.fragment.module = fs;
  renderPipelineDesc.fragment.entryPoint = "main";

  WGpuColorTargetState colorTarget = WGPU_COLOR_TARGET_STATE_DEFAULT_INITIALIZER;
  colorTarget.format = config.format;
  renderPipelineDesc.fragment.numTargets = 1;
  renderPipelineDesc.fragment.targets = &colorTarget;

  renderPipeline = wgpu_device_create_render_pipeline(device, &renderPipelineDesc);
  assert(wgpu_is_render_pipeline(renderPipeline));

  emscripten_request_animation_frame_loop(raf, 0);
}

void ObtainedWebGpuAdapter(WGpuAdapter result, void *userData)
{
  assert(userData == (void*)42);
  assert(wgpu_is_adapter(result));

  adapter = result;

  WGpuSupportedLimits limits;
  WGPU_FEATURES_BITFIELD features = wgpu_adapter_get_features(adapter);
  wgpu_adapter_get_limits(adapter, &limits);

#define TEST_FEATURE(x) emscripten_mini_stdio_printf("Adapter supports feature " #x ": %s\n", (features & (x) ? "yes" : "no"))
  TEST_FEATURE(WGPU_FEATURE_DEPTH_CLIP_CONTROL);
  TEST_FEATURE(WGPU_FEATURE_DEPTH32FLOAT_STENCIL8);
  TEST_FEATURE(WGPU_FEATURE_TEXTURE_COMPRESSION_BC);
  TEST_FEATURE(WGPU_FEATURE_TEXTURE_COMPRESSION_ETC2);
  TEST_FEATURE(WGPU_FEATURE_TEXTURE_COMPRESSION_ASTC);
  TEST_FEATURE(WGPU_FEATURE_TIMESTAMP_QUERY);
  TEST_FEATURE(WGPU_FEATURE_INDIRECT_FIRST_INSTANCE);

#define ADAPTER_LIMIT(x) emscripten_mini_stdio_printf("Adapter limit " #x ": %u\n", limits. x);
  ADAPTER_LIMIT(maxTextureDimension1D);
  ADAPTER_LIMIT(maxTextureDimension2D);
  ADAPTER_LIMIT(maxTextureDimension3D);
  ADAPTER_LIMIT(maxTextureArrayLayers);
  ADAPTER_LIMIT(maxBindGroups);
  ADAPTER_LIMIT(maxDynamicUniformBuffersPerPipelineLayout);
  ADAPTER_LIMIT(maxDynamicStorageBuffersPerPipelineLayout);
  ADAPTER_LIMIT(maxSampledTexturesPerShaderStage);
  ADAPTER_LIMIT(maxSamplersPerShaderStage);
  ADAPTER_LIMIT(maxStorageBuffersPerShaderStage);
  ADAPTER_LIMIT(maxStorageTexturesPerShaderStage);
  ADAPTER_LIMIT(maxUniformBuffersPerShaderStage);
  ADAPTER_LIMIT(minUniformBufferOffsetAlignment);
  ADAPTER_LIMIT(minStorageBufferOffsetAlignment);
  ADAPTER_LIMIT(maxVertexBuffers);
  ADAPTER_LIMIT(maxBufferSize);
  ADAPTER_LIMIT(maxVertexAttributes);
  ADAPTER_LIMIT(maxVertexBufferArrayStride);
  ADAPTER_LIMIT(maxInterStageShaderComponents);
  ADAPTER_LIMIT(maxComputeWorkgroupStorageSize);
  ADAPTER_LIMIT(maxComputeInvocationsPerWorkgroup);
  ADAPTER_LIMIT(maxComputeWorkgroupSizeX);
  ADAPTER_LIMIT(maxComputeWorkgroupSizeY);
  ADAPTER_LIMIT(maxComputeWorkgroupSizeZ);
  ADAPTER_LIMIT(maxUniformBufferBindingSize);
  ADAPTER_LIMIT(maxStorageBufferBindingSize);

  WGpuDeviceDescriptor deviceDesc = {};
  wgpu_adapter_request_device_async(adapter, &deviceDesc, ObtainedWebGpuDevice, (void*)43);
}

int main()
{
  WGpuRequestAdapterOptions options = {};
  navigator_gpu_request_adapter_async(&options, ObtainedWebGpuAdapter, (void*)42);
}