On Fri Nov 17 09:38:11 2023 +0000, Giovanni Mascellani wrote:
I don't think I agree with this. I'd reject valid programs rather than miscompile invalid programs. I think it makes sense to leave your test as `todo` for now, if you (understandably) don't want to work out all the overloading rules for this.
Okay, I shouldn't have relied on my memory alone. Also with your suggestion to change the test from `uint4` to `float4` we now have a test that shows the importance of 4/5, so I am okay with dropping 5/5.
I will keep in mind to gather some tests for my assessment that the exact matches always have priority when there are multiple compatible function overloads, or maybe I can be easily proven wrong with a counter example, but that can be left to another MR.