[Mlir-commits] [mlir] [MLIR][OpenMP] Move loop wrapper verification to the interface (NFC) (PR #110505)

[Sergio Afonso]

skatrak wrote:

> If the second operation is always `omp.terminator`, would it make sense to make the terminator implicit?

My first approach was indeed making these operations `SingleBlockImplicitTerminator`, but I replaced that back into `SingleBlock` because making the terminator optional (i.e. allowing it both to be specified and not specified) creates some complexity for the verifier.

I guess now that things are a bit more fleshed out I could look again into this. Maybe after landing the PR stack to unify the handling of block argument-defining ops, because I seem to remember there was some interaction between custom parsing of the region and the implicit terminators. Let me know if that works for you.

https://github.com/llvm/llvm-project/pull/110505