[Mlir-commits] [mlir] [MLIR][OpenMP] Group clause operands into structures (PR #86797)

[Sergio Afonso]

skatrak wrote:

> Could you explain a bit more about what you mean by the order of clauses here? We have `oilist` (https://mlir.llvm.org/docs/DefiningDialects/Operations/) in the assembly format to make it order-independent. I guess the usage of block arguments has made this a bit difficult. Is that the concern or am I on the wrong track here?

Yes, `oilist` allows specifying the clauses in any order in the textual MLIR representation. However, the C++ constructors of these MLIR operations do expect that arguments are passed in the same order they appear in the operation definition.

Given the relatively large amount of clauses some operations have, this has become a bit error-prone and that's what the new structure-based constructors address. These can be seen in use, together with other changes to use the new structures, in the next PR in the stack:  #86802.

https://github.com/llvm/llvm-project/pull/86797