Andreas Süßenbach

I've conducted a couple of compile-time experiments now, and conclude a couple of things: - There seems to be no single compile-time killer feature that dominates or at least contributes...

@RYDB3RG Thanks a lot for your investigations! Your results imply, that it' not a reasonable approach to separate the vulkan.hpp headers into a set of interface- and a set of...

As a side note: using vulkan.hpp (or vulkan_raii.hpp) as a precompiled header substantially reduces compile time.

Would you mind to provide a PR with a change that runs with your environment?

You're right, this is the main drawback of the vk::raii-classes: you can't get rid of all DescriptorSets of a DescriptorPool with just one call (and equally for CommandBuffers and CommandPool)....

As there seems to be no acceptable approach on how to handle "vulkan owned" objects here, and there's no mean to enforce correct usage, I'll close this issue for now....

That's an interesting approach for those who regularly build their vulkan.hpp on their own, instead of just using what's provided by the Vulkan SDK. Of course, when specifying an api-version,...

@Falkgaard You're absolutely right, the documentation on the new stuff in the namespace `vk::raii` is substantially less than satisfying. Be assured, that I'm working on some. Until then, let me...

With #1001, some kind of programming guide for the raii-wrapper classes has been added.

@amosnier The `vk::raii` classes provide a seperate set of handle wrapper classes. The new `vk::raii::Context` holds a `vk::DynamicLoader`, that handles the vulkan dll loading. That is, when using the `vk::raii`...