Linux compile broken on hl1 branch

[Rosentti]

Kind of a shame, wanted to try out some HL1 raytraced.
Arch Linux, RTX 2080.

Forced CLANG for the build.
(Tried to) build with:
cmake -DRG_WITH_NVIDIA_DLSS=OFF -DRG_WITH_SURFACE_WAYLAND=ON -DRG_WITH_SURFACE_XCB=OFF -DRG_WITH_SURFACE_XLIB=OFF -DRG_WITH_SURFACE_WIN32=OFF ..
Getting errors like:

[ 36%] Building CXX object CMakeFiles/RayTracedGL1.dir/Source/ASManager.cpp.o
In file included from /home/shared/Downloads/RayTracedGL1/Source/Swapchain.cpp:27:
/home/shared/Downloads/RayTracedGL1/Source/Utils.h:200:22: error: expected unqualified-id
    constexpr double M_PI = 3.1415926535897932384626433;
                     ^
/usr/include/math.h:1151:16: note: expanded from macro 'M_PI'
# define M_PI           3.14159265358979323846  /* pi */

make[2]: *** [CMakeFiles/RayTracedGL1.dir/build.make:146: CMakeFiles/RayTracedGL1.dir/Source/Scene.cpp.o] Error 1
/home/shared/Downloads/RayTracedGL1/Source/VulkanDevice_Init.cpp:99:13: error: incompatible pointer to integer conversion assigning to 'VkResult' from 'VkSurfaceKHR *' (aka 'VkSurfaceKHR_T **')
        r = ( instance, &wlInfo, nullptr, &surface );
            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

And some files reference , which doesn't actually seem to be a valid header (even on clang 15, where it should apparently exist).
Also, is the build 32-bit as that might explain some of these issues.

[Mis012]

I got it to compile with the following, but it seems 32-bit build is even more fun

diff --git a/Source/GltfExporter.cpp b/Source/GltfExporter.cpp
index 86073c7..4e3f350 100644
--- a/Source/GltfExporter.cpp
+++ b/Source/GltfExporter.cpp
@@ -40,6 +40,10 @@
 #include <span>
 #include <type_traits>
 
+#ifndef M_PI
+#define M_PI RTGL1::Utils::M_PI
+#endif
+
 namespace
 {
 template< class T >
@@ -901,7 +905,7 @@ private:
     static float LuminousFluxToCandela( float lumens )
     {
         // to lumens per steradian
-        return lumens / ( 4 * float( RTGL1::Utils::M_PI ) );
+        return lumens / ( 4 * (float)( M_PI ) );
     }
 
     static cgltf_light MakeLight( const RgDirectionalLightUploadInfo& sun )
diff --git a/Source/GltfImporter.cpp b/Source/GltfImporter.cpp
index b3658f4..59438c7 100644
--- a/Source/GltfImporter.cpp
+++ b/Source/GltfImporter.cpp
@@ -32,6 +32,10 @@
 
 #include <format>
 
+#ifndef M_PI
+#define M_PI Utils::M_PI
+#endif
+
 namespace RTGL1
 {
 namespace
@@ -910,7 +914,7 @@ void RTGL1::GltfImporter::UploadToScene( VkCommandBuffer           cmd,
 
         constexpr auto candelaToLuminousFlux = []( float lumensPerSteradian ) {
             // to lumens
-            return lumensPerSteradian * ( 4 * float( Utils::M_PI ) );
+            return lumensPerSteradian * ( 4 * (float)( M_PI ) );
         };
 
         auto makeExtras = []( const char* extradata ) {
diff --git a/Source/VulkanDevice_Init.cpp b/Source/VulkanDevice_Init.cpp
index d01a449..ce4cba4 100644
--- a/Source/VulkanDevice_Init.cpp
+++ b/Source/VulkanDevice_Init.cpp
@@ -96,7 +96,7 @@ VkSurfaceKHR GetSurfaceFromUser( VkInstance instance, const RgInstanceCreateInfo
         wlInfo.display                       = info.pWaylandSurfaceCreateInfo->display;
         wlInfo.surface                       = info.pWaylandSurfaceCreateInfo->surface;
 
-        r = ( instance, &wlInfo, nullptr, &surface );
+        r = vkCreateWaylandSurfaceKHR( instance, &wlInfo, nullptr, &surface );
         VK_CHECKERROR( r );
 
         return surface;

[Mis012]

ok, the following was needed for 32bit build, but who knows if it will work

diff --git a/Source/DebugWindows.cpp b/Source/DebugWindows.cpp
index 68a3360..bcee72a 100644
--- a/Source/DebugWindows.cpp
+++ b/Source/DebugWindows.cpp
@@ -254,7 +254,7 @@ RTGL1::DebugWindows::DebugWindows( VkInstance                               _ins
         .Device          = _device,
         .QueueFamily     = _queueFamiy,
         .Queue           = _queue,
-        .PipelineCache   = nullptr,
+        .PipelineCache   = (VkPipelineCache)nullptr,
         .DescriptorPool  = descPool,
         .Subpass         = 0,
         .MinImageCount   = swapchainImageCount,
diff --git a/Source/DepthCopying.cpp b/Source/DepthCopying.cpp
index 5ff0731..9df03fd 100644
--- a/Source/DepthCopying.cpp
+++ b/Source/DepthCopying.cpp
@@ -359,7 +359,7 @@ void RTGL1::DepthCopying::CreatePipeline( const ShaderManager* shaderManager )
         .basePipelineHandle  = VK_NULL_HANDLE,
     };
 
-    VkResult r = vkCreateGraphicsPipelines( device, nullptr, 1, &plInfo, nullptr, &pipeline );
+    VkResult r = vkCreateGraphicsPipelines( device, (VkPipelineCache)nullptr, 1, &plInfo, nullptr, &pipeline );
 
     VK_CHECKERROR( r );
     SET_DEBUG_NAME( device, pipeline, VK_OBJECT_TYPE_PIPELINE, "Rasterizer raster draw pipeline" );
diff --git a/Source/DrawFrameInfo.h b/Source/DrawFrameInfo.h
index 6a92784..33017b4 100644
--- a/Source/DrawFrameInfo.h
+++ b/Source/DrawFrameInfo.h
@@ -248,7 +248,7 @@ namespace detail
 
     inline void* ReadPNext( void* params )
     {
-        constexpr size_t pNextOffset = 8;
+        constexpr size_t pNextOffset = sizeof(void *);
 
         // clang-format off
         static_assert( offsetof( RgDrawFrameRenderResolutionParams, pNext ) == pNextOffset );
diff --git a/Source/LensFlares.cpp b/Source/LensFlares.cpp
index 932c96d..8b8ac66 100644
--- a/Source/LensFlares.cpp
+++ b/Source/LensFlares.cpp
@@ -554,13 +554,13 @@ void RTGL1::LensFlares::CreatePipelines( const ShaderManager* shaderManager )
     };
     info.stage.pSpecializationInfo = &spec;
 
-    VkResult r = vkCreateComputePipelines( device, nullptr, 1, &info, nullptr, &cullPipeline );
+    VkResult r = vkCreateComputePipelines( device, (VkPipelineCache)nullptr, 1, &info, nullptr, &cullPipeline );
     VK_CHECKERROR( r );
 }
 
 void RTGL1::LensFlares::DestroyPipelines()
 {
-    if( cullPipeline != nullptr )
+    if( cullPipeline != (VkPipeline)nullptr )
     {
         vkDestroyPipeline( device, cullPipeline, nullptr );
     }
diff --git a/Source/RTGL1.cpp b/Source/RTGL1.cpp
index bd6629e..6b570e7 100644
--- a/Source/RTGL1.cpp
+++ b/Source/RTGL1.cpp
@@ -25,10 +25,10 @@
 
 namespace
 {
-#define INITIALIZED_RGINSTANCE ( reinterpret_cast< RgInstance >( 1024 ) )
+#define INITIALIZED_RGINSTANCE ( reinterpret_cast< RgInstance >( (RgInstance)1024 ) )
 
 
-RgInstance                             g_deviceRgInstance{ nullptr };
+RgInstance                             g_deviceRgInstance{ (RgInstance)nullptr };
 std::unique_ptr< RTGL1::VulkanDevice > g_device{};
 
 RTGL1::VulkanDevice* TryGetDevice( RgInstance rgInstance )
@@ -63,7 +63,7 @@ RgMessageSeverityFlags g_printSeverity{ 0 };
 
 RgResult rgCreateInstance( const RgInstanceCreateInfo* pInfo, RgInstance* pResult )
 {
-    *pResult = nullptr;
+    *pResult = (RgInstance)nullptr;
 
     if( TryGetDevice( g_deviceRgInstance ) )
     {
@@ -130,7 +130,7 @@ RgResult rgDestroyInstance( RgInstance rgInstance )
     try
     {
         g_device.reset();
-        g_deviceRgInstance = nullptr;
+        g_deviceRgInstance = (RgInstance)nullptr;
     }
     catch( RTGL1::RgException& e )
     {

[sultim-t]

Oof, sorry, but ray tracing doesn't work in 32-bit :(

[Mis012]

are you sure? it didn't even compile in 32bit without the patch...
I get the menu to display now, but when I try to launch New Game, the whole system locks up...
is there some way to load a specific (ideally small) map directly from cmdline?

[sultim-t]

There's no support for 32-bit applications on the driver side for technical reasons, I assume. That is why applications need to be translated to 64-bit :(

[Mis012]

sounds like a windows-only problem, and this issue is specifically about Linux

[Mis012]

well, maybe for Nvidia proprietary drivers it would be an issue on Linux as well, but not for Mesa