[Mlir-commits] [mlir] [mlir][arith] doc updates for ub semantics, and int representations (PR #72932)

[Fehr Mathieu]

math-fehr wrote:

Just for information, I currently encoded these semantics in my [SMT dialect project](https://github.com/opencompl/xdsl-smt) (the project is still WIP, and I plan on moving it to C++).
With these current semantics, I fuzzed a bit MLIR to find if MLIR is currently respecting these semantics in the `canonicalize` and `arith-expand` passes.
For now, I only found a single bug, which is the canonicalization of `select`, which is more poisonous if we were to chose these semantics. I can report it if we decide to agree on these semantics.

Otherwise, I plan in the next few days on encoding all the `canonicalize`, `arith-expand`, and later `convert-arith-to-llvm` rewrite patterns to my tool, a bit Alive-style, so we could know for sure if MLIR is currently respecting these poison semantics or not. I guess that *could* help making sure whatever semantics we chose would be respected.

https://github.com/llvm/llvm-project/pull/72932