[PATCH] D14265: DI: Reverse direction of subprogram -> function edge.

[Adrian Prantl via llvm-commits]

aprantl added a subscriber: aprantl.
aprantl added a comment.

I'm impressed that we apparently never needed the other direction.
This looks good from a debug info point of view.

Two questions/remarks:

- Could you include your upgrade script so people can upgrade their out-of-tree testcases?
- Shouldn't we drop the function field from the DISubprogram bitcode representation?

thanks for working on this!
adrian


http://reviews.llvm.org/D14265