Hi David, Thanks for reviewing the change. I might need some clarifications. I tried the sample codes from the stackoverflow link (thanks for doing the research!), and it appears that we do not implement this bad behavior from MSVC even with "clang -cc1 -triple i686-pc-win32 -fms-extensions". So I deduce that "clang -fms-extensions" will behave differently than "cl /Ze" on Windows. On the other hand, this proposed patch is only making -fms-extensions behave the same on Windows and non-Windows targets. Since _MSC_EXTENSIONS is currently defined on Windows target, I guess programmers will already be surprised if they are using this macro to test MSVC-specific behaviors? - Gao.
http://llvm-reviews.chandlerc.com/D3034 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
