[PATCH] D17241: [Polly] [Refactor] Move isl_ctx into Scop.

[Tobias Grosser via llvm-commits]

grosser added a comment.

Thanks ether for explaining these things in detail. It seems indeed to be the case that we free passes in a rather unpredictable order. If this is the case, the shared pointer approach seems OK and I now convinced myself that it works and does not prevent us from loosing the ability to understand if we leak memory.

However, if we could understand why this freeing behavior is so unpredictable that would be nice.

Best,
Tobias


http://reviews.llvm.org/D17241