[PATCH] D73360: [OpenCL] Restrict address space conversions in nested pointers

[Anastasia Stulova via Phabricator via cfe-commits]

Anastasia added a comment.

In D73360#1850001 <https://reviews.llvm.org/D73360#1850001>, @rjmccall wrote:

> Is there no follow-up code when actually emitting the failure diagnostic which tries to figure out a more specific cause of failure?


After looking at this a bit more it seems C++ part calls the diagnostic code earlier:

  Sema::PerformImplicitConversion:
  
     case ImplicitConversionSequence::BadConversion:
       bool Diagnosed =
           DiagnoseAssignmentResult(Incompatible, From->getExprLoc(), ToType,
                                    From->getType(), From, Action);

It just seems to always pass `Incompatible`. We could add follow up code before? This would be easier than modifying the existing flow completely. However, looking at this holistically we would need to repeat all checks classifying the failure types...


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D73360/new/

https://reviews.llvm.org/D73360