Missing Abstract Int/Float preconditions for Conversion Expressions #2788

Missing Abstract Int/Float preconditions for Conversion Expressions

bool

• Abstract Int/Float -> bool (table entries could be reduced to 2 and preconditions changed to say AbstractInt, i32 or u32 and AbstractFloat, f32 or f16)

  

i32

• AbstractInt -> i32

• AbstractFloat -> i32 (table entries could be reduced to 1 and preconditions changed to say AbstractFloat, f32 or f16)

  

u32

• AbstractInt -> u32 (I think the table entry for this is wrong since it says "Reinterpretation of bits" and AbstractInt's are essentially i64's)

  

• AbstractFloat -> u32 (again, table entries could be reduced to 1 and preconditions changed to say AbstractFloat, f32 or f16)

  

f32 and f16

• AbstractInt -> f32 / f16 (again, table entries could be reduced to 1 for each and preconditions changed to say AbstractInt, i32 or u32)
• AbstractFloat -> f32 / f16

The list above is concerned with "Scalar conversion type rules" but the same adjustments could be done for the "Vector conversion type rules" and "Matrix conversion type rules" sections.