[Lldb-commits] [PATCH] D15326: Rework breakpoint language filtering to use the symbol context's language.

Jim Ingham via lldb-commits lldb-commits at lists.llvm.org
Wed Dec 9 10:46:44 PST 2015


jingham added a comment.

"cstring_is_mangled" does NOT mean that the string is a C++ mangled string, it means it is ANY kind of mangled string - you can tell because it is used to determine whether to populate the m_mangled and m_demangled components of the Mangled structure, which it should do regardless of the language the compiler happens to be mangling .  If you look at the Github sources with Swift support, you will see that this returns true for both C++ mangled names and Swift mangled names, as it should.

It looks like you are using a function that had this error originally, so it is not your fault per se.  We were probably getting away with it because it was never used in a case where its result mattered (it isn't used much...)  But you are now using it somewhere where the incorrect usage would matter.  So it needs to be fixed before the patch is correct.

For the nonce, it is probably good enough for Mangled::GuessLanguage to return C++ if the mangling scheme is Itanium or MSVC.  We definitely have to figure out a better way to this, so a comment to that effect would also be in order.


Repository:
  rL LLVM

http://reviews.llvm.org/D15326





More information about the lldb-commits mailing list