On Tue, Aug 18, 2009 at 7:01 AM, Anton Korobeynikov<[email protected]> wrote: > Hello, John > >> Adds _M_IX86 for the Windows target. > That's not correct. And strictly speaking current windows target is > not correct as well. This define (and _MSC_VER defined already) are > vcpp-specific. They should not be defined neither for mingw (which > uses current windows target) nor for generic windows target. Basically > we need to have separate "VCPPCompatible" target which will derive > from windows target and define vcpp-specific defines, hacks and > quirks.
clang does not define _MSC_VER; it never has, and never will unless we introduce a vcpp emulation mode, which I don't find particularly likely. Some quick web research shows _M_IX86 does not necessarily imply MSVC. Apparently Borland defines it, and the MinGW windows.h defines it if the compiler doesn't. I don't see any issue with defining it. I'm not sure what the distinction between mingw and a "generic" windows target is in this context. And we currently default to -fms-extensions on Windows, but it can be explicitly disabled. -Eli _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
