5 Jul
2023
5 Jul
'23
9:23 a.m.
At first sight this doesn't look wrong, but previous tests (see `tests/hlsl/majority_*.shader_test`) show that the way majority interacts with `pragma`'s and `typedef`'s is quite complicated. In particular, there are some subtle differences between having column majority as a result of setting it explicitly via a `pragma` and as a result of not having set any other default. How that interacts with the majority set via a compilation flag is therefore not necessarily obvious (i.e., is explicitly setting a column majority via a compilation flag equivalent to setting it via `pragma`, or to leaving it to the unset default?), and for that reason it would be particularly appropriate to have tests for that.