On Sat, 12 Dec 2020 at 01:57, Zebediah Figura (she/her) zfigura@codeweavers.com wrote:
On 12/11/20 12:11 PM, Henri Verbeet wrote:
Is reusing enum vkd3d_shader_error for warnings instead of introducing a separate enumeration here intentional?
It was intentional, though of course not necessary. I guess my reasoning was that there's no reason to let any numerical values coincide, and hence no reason to bother separating the enumerations...
Fair enough, although it does look a bit odd to pass an enum vkd3d_shader_error to vkd3d_shader_warning(). We may want to consider starting the preprocessor warning range at e.g. 4200 or 4300 in that case though; if we were to add VKD3D_SHADER_LOG_WARNING and VKD3D_SHADER_LOG_INFO ranges for all the existing VKD3D_SHADER_LOG_ERROR ranges using the current allocation scheme, 4-digit error codes would already be insufficient. (And there are still some sources/targets we may potentially add; HLSL of course, but also GLSL, perhaps MSL, probably DXIL, D3D shader model 1-3 bytecode.)