Backends: WebGPU: Passing explicit buffer sizes as validation layers appears to not do what the in-flux specs says. (#4766

backends/imgui_impl_sdlrenderer.cpp

@@ -1,14 +1,17 @@
 // dear imgui: Renderer Backend for SDL_Renderer
 // (Requires: SDL 2.0.17+)
 
-// Important to understand: SDL_Renderer is an _optional_ component of SDL. We do not recommend you use SDL_Renderer
-// because it provide a rather limited API to the end-user. We provide this backend for the sake of completeness.
+// Important to understand: SDL_Renderer is an _optional_ component of SDL.
 // For a multi-platform app consider using e.g. SDL+DirectX on Windows and SDL+OpenGL on Linux/OSX.
+// If your application will want to render any non trivial amount of graphics other than UI,
+// please be aware that SDL_Renderer offers a limited graphic API to the end-user and it might
+// be difficult to step out of those boundaries.
+// However, we understand it is a convenient choice to get an app started easily.
 
 // Implemented features:
 //  [X] Renderer: User texture binding. Use 'SDL_Texture*' as ImTextureID. Read the FAQ about ImTextureID!
 // Missing features:
-//  [ ] Renderer: Support for large meshes (64k+ vertices) with 16-bit indices.
+//  [ ] Renderer: Support for large meshes (64k+ vertices) with 16-bit indices (SDL_RenderGeometryRaw() is missing a vertex offset).
 
 // You can copy and use unmodified imgui_impl_* files in your project. See examples/ folder for examples of using this.
 // If you are new to Dear ImGui, read documentation from the docs/ folder + read the top of imgui.cpp.

backends/imgui_impl_sdlrenderer.h

@@ -1,14 +1,17 @@
 // dear imgui: Renderer Backend for SDL_Renderer
 // (Requires: SDL 2.0.17+)
 
-// Important to understand: SDL_Renderer is an _optional_ component of SDL. We do not recommend you use SDL_Renderer
-// because it provide a rather limited API to the end-user. We provide this backend for the sake of completeness.
+// Important to understand: SDL_Renderer is an _optional_ component of SDL.
 // For a multi-platform app consider using e.g. SDL+DirectX on Windows and SDL+OpenGL on Linux/OSX.
+// If your application will want to render any non trivial amount of graphics other than UI,
+// please be aware that SDL_Renderer offers a limited graphic API to the end-user and it might
+// be difficult to step out of those boundaries.
+// However, we understand it is a convenient choice to get an app started easily.
 
 // Implemented features:
 //  [X] Renderer: User texture binding. Use 'SDL_Texture*' as ImTextureID. Read the FAQ about ImTextureID!
 // Missing features:
-//  [ ] Renderer: Support for large meshes (64k+ vertices) with 16-bit indices.
+//  [ ] Renderer: Support for large meshes (64k+ vertices) with 16-bit indices (SDL_RenderGeometryRaw() is missing a vertex offset).
 
 #pragma once
 #include "imgui.h"      // IMGUI_IMPL_API

backends/imgui_impl_wgpu.cpp

@@ -13,6 +13,7 @@
 
 // CHANGELOG
 // (minor and older changes stripped away, please see git history for details)
+//  2021-11-29: Passing explicit buffer sizes to wgpuRenderPassEncoderSetVertexBuffer()/wgpuRenderPassEncoderSetIndexBuffer().
 //  2021-08-24: Fix for latest specs.
 //  2021-05-24: Add support for draw_data->FramebufferScale.
 //  2021-05-19: Replaced direct access to ImDrawCmd::TextureId with a call to ImDrawCmd::GetTexID(). (will become a requirement)
@@ -313,8 +314,8 @@ static void ImGui_ImplWGPU_SetupRenderState(ImDrawData* draw_data, WGPURenderPas
     wgpuRenderPassEncoderSetViewport(ctx, 0, 0, draw_data->FramebufferScale.x * draw_data->DisplaySize.x, draw_data->FramebufferScale.y * draw_data->DisplaySize.y, 0, 1);
 
     // Bind shader and vertex buffers
-    wgpuRenderPassEncoderSetVertexBuffer(ctx, 0, fr->VertexBuffer, 0, 0);
-    wgpuRenderPassEncoderSetIndexBuffer(ctx, fr->IndexBuffer, sizeof(ImDrawIdx) == 2 ? WGPUIndexFormat_Uint16 : WGPUIndexFormat_Uint32, 0, 0);
+    wgpuRenderPassEncoderSetVertexBuffer(ctx, 0, fr->VertexBuffer, 0, fr->VertexBufferSize);
+    wgpuRenderPassEncoderSetIndexBuffer(ctx, fr->IndexBuffer, sizeof(ImDrawIdx) == 2 ? WGPUIndexFormat_Uint16 : WGPUIndexFormat_Uint32, 0, fr->IndexBufferSize);
     wgpuRenderPassEncoderSetPipeline(ctx, g_pipelineState);
     wgpuRenderPassEncoderSetBindGroup(ctx, 0, g_resources.CommonBindGroup, 0, NULL);
 

docs/BACKENDS.md

@@ -128,7 +128,10 @@ Once it works, if you really need it you can replace parts of backends with your
 and you have high-level systems everywhere.<BR>
 Suggestion: try using a non-portable backend first (e.g. win32 + underlying graphics API) to get
 your desktop builds working first. This will get you running faster and get your acquainted with
-how Dear ImGui works and is setup. You can then rewrite a custom backend using your own engine API.
+how Dear ImGui works and is setup. You can then rewrite a custom backend using your own engine API...
+
+Generally:
+It is unlikely you will add value to your project by creating your own backend.
 
 Also:
 The [multi-viewports feature](https://github.com/ocornut/imgui/issues/1542) of the 'docking' branch allows

docs/CHANGELOG.txt

@@ -70,13 +70,16 @@ Other Changes:
   (an additional ItemSpacing.y was declared, affecting scrollbar range).
 - Drag and Drop: BeginDragDropSource() with ImGuiDragDropFlags_SourceAllowNullID doesn't lose
   tooltip when scrolling. (#143)
-- Backends: Vulkan: Call vkCmdSetScissor() at the end of render with a full-viewport to reduce
-  likehood of issues with people using VK_DYNAMIC_STATE_SCISSOR in their app without calling
-  vkCmdSetScissor() explicitly every frame. (#4644)
 - Misc: Added missing ImGuiMouseCursor_NotAllowed cursor for software rendering (when the
   io.MouseDrawCursor flag is enabled). (#4713) [@nobody-special666]
 - Misc: Fix MinGW DLL build issue (when IMGUI_API is defined). [@rokups]
 - CI: Add MinGW DLL build to test suite. [@rokups]
+- Backends: Vulkan: Call vkCmdSetScissor() at the end of render with a full-viewport to reduce
+  likehood of issues with people using VK_DYNAMIC_STATE_SCISSOR in their app without calling
+  vkCmdSetScissor() explicitly every frame. (#4644)
+- Backends: WebGPU: Passing explicit buffer sizes to wgpuRenderPassEncoderSetVertexBuffer() and
+  wgpuRenderPassEncoderSetIndexBuffer() functions as validation layers appears to not do what the
+  in-flux specs says. (#4766) [@meshula]
 
 
 -----------------------------------------------------------------------