[PATCH] D52581: [AST] Revert mangling changes from r339428

[David Chisnall via Phabricator via cfe-commits]

theraven added a comment.

> I would have done the same for the GNUstep RTTI here, except I don't actually
>  see the code for that anywhere, and no tests seem to break either, so I
>  believe it's not upstreamed yet.

I'm not sure I understand this comment.  The compiler code is in LLVM and the tests that you've modified are the ones that would fail.  The corresponding code is upstreamed in the runtime, to generate EH metadata for the throw with the same mangling.  If you are going to change the mangling, please make sure that the existing mangling is preserved for EH when compiling for the GNUstep / Microsoft ABI on Windows.

I'm also deeply unconvinced by the idea that it's okay to have a `struct I` in one compilation unit and a `@class I` in another.  The `struct` version will not do any of the correct things with respect to memory management.  Code that has this idiom is very likely to be broken.


Repository:
  rC Clang

https://reviews.llvm.org/D52581