[Mlir-commits] [mlir] Fix int range analysis (PR #192235)

[Ivan Butygin]

Hardcode84 wrote:

Ok, I was thinking of something along the line "fallback to conservative value after a some predefined number of meet/join attempts on lattice element" in base dataflow infra, but I don't have a concrete design yet and also, we need to avoid introducing too much overhead for all dataflow analyses.

https://github.com/llvm/llvm-project/pull/192235