Martin Ejdestig

Sorry for the many updates. I should have checked `.travis.yml` earlier to see what was actually done when checking pr, would have minimized updates a bit. All tests should pass...

Rebased to latest master.

Sure, I can have a look. I just checked quickly and wonder a bit if using `SIGUSR1` is safe. E.g. if the application uses `SIGUSR1` for something it can still...

- I think it would be better if a library does not "occupy" `SIGUSR1`. Nothing for application authors to care about. No documentation to read. :) And what do you...

OK. Do you want me to close this pr or do you want me to rebase it after https://github.com/GENIVI/dlt-daemon/commit/fae6bf7608adcf61f71c3240d2e96c8f64135642 has been merged?

I am thinking that it perhaps makes sense to remove use of `pthread_cancel()` (and `SIGUSR1`) and align thread cleanup on all platforms after a release has been done or something...

Hello @SaschaWillems , I was just looking at your shadow mapping example on the master branch and wondered why it was enough with a single shadow map buffer. Found `vkQueueWaitIdle()`...

I think that is all there is to it, pImmutableSamplers just needs to not be set. From https://registry.khronos.org/vulkan/specs/1.3-extensions/man/html/VkDescriptorSetLayoutBinding.html : > pImmutableSamplers affects initialization of samplers. If descriptorType specifies a VK_DESCRIPTOR_TYPE_SAMPLER...

Well, I do not care personally. (And I no longer work for company for which patch was made.) Feel free to rebase and merge yourself. Or close and keep hogging...

I think I got bit by this. Investigated a slowdown related to vertex/index buffers that are updated every frame from the CPU. Bisecting lead me to commit that integrated VMA....