[Mlir-commits] [mlir] [mlir][LLVMIR] Skip or and shl folding for mismatched constant types (PR #226026)

[Tobias Gysi]

gysit wrote:

> Would you prefer changing test.constant and its dependent tests instead?

That is probably not possibly since the operation is specifically used to test some arith folders/patterns that apparently check the attribute type. 

That means the options are to stick to status quo and push more towards enforcing type equality on the constant operations themselves (or possibly in the folding infra), or we say there is prior art and we land your pr to strengthen the checks in the LLVM dialect folders.

Since there is prior art in Arith, I am ok with landing this pr. @joker-eph what is your take on this? 


https://github.com/llvm/llvm-project/pull/226026