Question about instance flags validation and debug

[almarklein]

Recently the WGPUInstanceExtras struct was added, enabling setting WGPUInstanceFlag_Debug and WGPUInstanceFlag_Validation. These directly map to wgpu-types flags:

wgpu-native/src/conv.rs

Lines 260 to 272 in fcc480b

	pub fn map_instance_flags(flags: native::WGPUInstanceFlag) -> wgt::InstanceFlags {
	let mut result = wgt::InstanceFlags::empty();
	if (flags & native::WGPUInstanceFlag_Debug) != 0 {
	result.insert(wgt::InstanceFlags::DEBUG);
	}
	if (flags & native::WGPUInstanceFlag_Validation) != 0 {
	result.insert(wgt::InstanceFlags::VALIDATION);
	}
	if (flags & native::WGPUInstanceFlag_DiscardHalLabels) != 0 {
	result.insert(wgt::InstanceFlags::DISCARD_HAL_LABELS);
	}
	result
	}

Questions:

How are things different when (either of) these flags is enabled?

Is the validation flag related to Vulkan validation layers or is the validation specific to wgpu?

How does the debug flag relate to wgpuSetLogLevel(WGPULogLevel_Debug)?

[cwfitzgerald]

Is the validation flag related to Vulkan validation layers or is the validation specific to wgpu?

vulkan/dx12 validation layers. It probably should be renamed to like "underlying validation".

How does the debug flag relate to wgpuSetLogLevel(WGPULogLevel_Debug)?

It doesn't, this is about debug information in shaders and such.

[almarklein]

Thanks for the answers. Sorry for the many questions. This looks like something that can be a handy debugging tool, but I still don't understand what to expect 😊

It says here "Enable validation, if possible". Under what conditions is it possible, and how do I know that its enabled?

How is this related to enabling the Vulkan validation layers via the Lunar SDK (by opening the Vulkan Configurator)?

If more validation is done, what effect does this have in practice? Will it be more strict and fail if these stricter conditions are not met, or is there more information written to the logger?