On Sep 5, 2012, at 1:59 AM, Enea Zaffanella wrote: > On 09/04/2012 07:47 PM, John McCall wrote: >> On Sep 2, 2012, at 11:02 PM, Enea Zaffanella wrote: >>> Resending to cfe-commit. >>> >>> On 08/31/2012 12:09 PM, Enea Zaffanella wrote: >>>> Hello. >>>> >>>> According to language standards, the wchar_t type is >>>> - a library provided type in C >>>> - a native type in C++. >>>> >>>> There are C++ compilers allowing users to enable/disable the native >>>> support. For instance, MSVC has option /Zc:wchar_t. Even though the >>>> default in newer versions of MSVC is to enable wchar_t support, there is >>>> much code in the real world that will only work if this is disabled >>>> (e.g., Qt X.Y.Z with X < 5). >>>> >>>> The attached simple patch adds flag WChar to LangOptions. >>>> The flag is similar to Bool: if set, the native wchar_t type is >>>> supported (i.e., wchar_t will be lexed as a keyword), otherwise it won't >>>> be supported (i.e., it will be a normal identifier). Currently, the flag >>>> is initialized the same as CPlusPlus, so that the addition should cause >>>> no visible change. >>>> >>>> The addition of this flag will be enough for people using clang as a set >>>> of libraries. For more "conventional" uses, it might be worth >>>> considering the addition of a suitable command line option to toggle the >>>> flag. >>>> >>>> Opinions? >> >> Please add a -cc1 option to suppress native wchar_t; I think -fno-wchar >> would be sufficient. >> >> John. > > Here is the revised patch, including a testcase. > > OK to commit?
Looks great, thanks. John. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
