On Fri, 22 Feb 2019 at 02:51, Józef Kucia joseph.kucia@gmail.com wrote:
On Thu, Feb 21, 2019 at 8:49 PM Henri Verbeet hverbeet@gmail.com wrote:
On Thu, 21 Feb 2019 at 15:10, Józef Kucia joseph.kucia@gmail.com wrote:
Allows running all tests cleanly on Intel and Nvidia. It might be a controversial change, but it should make development slightly easier.
I don't mind the concept (although in terms of terminology, I think I'd prefer something along the lines of "broken_if()" to distinguish it from actual vkd3d todo's), but doesn't this mean these will start failing if those bugs are ever fixed?
Yes, these will start failing when driver bugs are fixed. We could guard these conditions with an env var or a command line option. Currently, it should be possible to disable those conditions by setting VKD3D_DISABLE_EXTENSIONS=VK_KHR_driver_properties. Other option is to make them special and count them as broken when a test fails, and treat them as success when a test passes.
I think we essentially want to just ignore the result if the condition is true, yes.
I agree that a separate name to mark driver-specific test failures could be a good idea. Unfortunately, broken() is taken. It's used for conditions which are accepted only on Windows. I'm not sure how to name them. A few random ideas: broken_driver_if(), driver_bug_if().
I don't have any great naming suggestions either.