[Mlir-commits] [mlir] [MLIR][Arith][Vector] Reject i0 integer type in arith and vector ops (PR #183589)

[Jakub Kuderski]

kuhar wrote:

> @kuhar : is this what you had in mind? This may deserve an RFC though.

Yes, exactly. I strongly prefer this over band aid patches that fix non-zero bitwdith assumptions scattered across the codebase. 

I'm not sure if an RFC is necessary -- I think i0 was allowed unintentionally after most of these dialects was already in place and we missed ODS constraints. I'm not aware of anyone using these, but if you want to create one, it should be pretty straightforward.

https://github.com/llvm/llvm-project/pull/183589