[flang-commits] [flang] [fir] Support promoting `fir.do_loop` with results to `affine.for`. (PR #137790)

[via flang-commits]

NexMing wrote:

> Thank you for contributing this. After this patch, do you know of any remaining limitations of the affine promotion pass?
> 
> And if you don't mind me asking, what are you using this for?

I haven't done detailed testing, so I'm not sure how many features are still missing.
Currently, I'm planning to promote to Affine for some advanced optimizations. I've already successfully tried one optimization path: `FIR → Affine (Optimization) → SCF → OMP → LLVM`, to automatically apply OpenMP parallelization on loops.

https://github.com/llvm/llvm-project/pull/137790