Lesson%201%3A%20Hello%20Metal/2.%20Hello%20Triangle/

[giscus[bot]]

Lesson%201%3A%20Hello%20Metal/2.%20Hello%20Triangle/

https://metaltutorial.com/Lesson%201%3A%20Hello%20Metal/2.%20Hello%20Triangle/

[romanWSgit]

Great Tutorial so far,

but...

Maybe I forgot to import some Library or something, but "simd::float3" is marked as an undeclared identifier in
void MTLEngine::createTriangle()
{
simd::float3 triangleVertices[] =
{
{-0.5f, -0.5f, 0.0f},
{ 0.5f, -0.5f, 0.0f},
{ 0.0f, 0.5f, 0.0f}
};

triangleVertexBuffer = metalDevice->newBuffer(&triangleVertices,
sizeof(triangleVertices),
MTL::ResourceStorageModeShared);
}

wich is in located in mtl_engine.mm in my case.

Kind of stuck here...
Help would be appreciated!

Regards, Roman

[romanWSgit]

I included "#include <simd/simd.h>" in mil_engine.mm and this seems to work. Maybe I overlooked this.
Nevertheless ...great tutorial!

[wmarti]

Nope, seems like I actually overlooked that! Nice catch :). I'll update the docs right now. Just note that the site might go offline for a minute or so when I do, GitHub pages likes to mess everything up...

[wmarti]

Docs have been updated! Do let me know if you find anything else I've missed, I'll be sure to fix it right away :)

[romanWSgit]

Thanks very much for the fast reply.

[MERL10N]

With regards to the objective-c code needed in MTLEngine, you can abstract them away so that MTLEngine and main can be .cpp instead of .mm

First, replace @autoreleasepool in the run() function with NS::AutoReleasePool* pPool
(You can declare it in the run function or in MTLEngine.hpp as I did)

void MTLEngine::run()
{
while (!glfwWindowShouldClose(glfwWindow)) {
pPool = NS::AutoreleasePool::alloc()->init();
metalDrawable = layer->nextDrawable();
draw();
pPool->release();
glfwPollEvents();
}
}

Note that you can make this change in main.cpp as well

Next, declare in MTLEngine.hpp

CA::MetalLayer* layer;

and add these lines in the init() function. Don't compile yet because there will still be errors.

layer = CA::MetalLayer::layer();
layer->setDevice(metalDevice);
layer->setPixelFormat(MTL::PixelFormatBGRA8Unorm);
layer->setDrawableSize(CGSizeMake(width, height));
GLFWBridge::AddLayerToWindow(glfwWindow, layer);

Thirdly,
remove these in MTLEngine.hpp

#include <Metal/Metal.h>
#include <GLFW/glfw3native.h>
#define GLFW_EXPOSE_NATIVE_COCOA

#include <QuartzCore/CAMetalLayer.h>

then go ahead and include

#include <QuartzCore/CAMetalLayer.hpp>

Note that you should still keep #include <GLFW/glfw3.h> in MTLEngine

Lastly, this final step is based on what is found on stack overflow: https://stackoverflow.com/questions/76776751/how-to-use-metal-cpp-with-glfw

I named my file GLFWBridge.h/.mm

#ifndef GLFWBridge_h
#define GLFWBridge_h
struct GLFWwindow;

namespace CA
{
class MetalLayer;
}
namespace GLFWBridge
{
void AddLayerToWindow(GLFWwindow* window, CA::MetalLayer* layer);
}
#endif /* GLFWBridge_h */

#include "GLFWBridge.h"
#define GLFW_EXPOSE_NATIVE_COCOA
#include <GLFW/glfw3.h>
#include <GLFW/glfw3native.h>

#include <Metal/Metal.h>
#include <QuartzCore/QuartzCore.h>

namespace GLFWBridge
{
void AddLayerToWindow(GLFWwindow* window, CA::MetalLayer* layer)
{
NSWindow* cocoa_window = glfwGetCocoaWindow(window);
CAMetalLayer* native_layer = (__bridge CAMetalLayer*)layer;
[[cocoa_window contentView] setLayer:native_layer];
[native_layer setMaximumDrawableCount:2];
[[cocoa_window contentView] setWantsLayer:YES];
[[cocoa_window contentView] setNeedsLayout:YES];
}
}

Back in MTLEngine.mm

Go ahead and add this

#include "GLFWBridge.h"

and this line in init() after layer->setDrawableSize(CGSizeMake(width, height));

GLFWBridge::AddLayerToWindow(glfwWindow, layer);

After all these, you can go ahead and switch MTLEngine.mm to MTLEngine.cpp, as well as main.mm to main.cpp.

[MissBidule]

FYI, I did your solution and it worked great... except at the 3D step, it made my cube turn all laggy so I had to delete this line : [native_layer setMaximumDrawableCount:2] as its default value is 3.
I tried to find why it should be 3 and not 2 but did not understand...

[kptr-juku]

Thanks, I was experimenting with replacing the MetalLayer with the CPP one myself, but I could not exactly 'bridge' it :)
I would add, that there is a typo in the first placement of NS::AutoReleasePool* pPool for the mtl_engine.h, it should be without capital 'R' - took me some time to figure this out :)
Also pixel format getter from layer should now be:
MTL::PixelFormat pixelFormat = (MTL::PixelFormat)layer->pixelFormat();

[AlexanderDevaikinEnscape]

Shouldn't we have released vertex and fragment shader objects because they were created using function with the new prefix in name?

vertexShader->release();
fragmentShader->release();

[shi-yan]

you forgot to release vertexShader and fragmentShader?

[chuzcjoe]

For anyone who might be interested in pure C++ implementation of this tutorial:
https://github.com/chuzcjoe/CORE/blob/master/examples/metal/MetalTriangleDemo/main.cpp