[LLVMbugs] [Bug 7858] New: a template type parameter name should hide a same-named member of that type

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Aug 9 23:55:34 PDT 2010


http://llvm.org/bugs/show_bug.cgi?id=7858

           Summary: a template type parameter name should hide a
                    same-named member of that type
           Product: clang
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: C++
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: zhanyong.wan at gmail.com
                CC: llvmbugs at cs.uiuc.edu, dgregor at apple.com


$ cat lookup_test.cc
template <typename T> void f(T* t) {
  t->T::foo();
}

struct S {
  void foo() {}
  enum T { SOME_CONSTANT };
};

int main() {
  S s;
  f(&s);
}

$ clang lookup_test.cc
lookup_test.cc:2:7: error: expected a class or namespace
  t->T::foo();
      ^
lookup_test.cc:12:3: note: in instantiation of function template specialization
'f<S>' requested here
  f(&s);
  ^

Clang seems to think that the name T in "t->T::foo()" resolves to S::T.  I tend
to think that it should be the template parameter T (and thus the code should
be valid).

I checked the C++03 standard 14.6 [temp.res] for the rules regarding name
resolution in template definitions.  Unfortunately I didn't find anything
conclusive.  The reasons for my belief that the code is valid are:

- Both GCC and Comeau accept the code.
- It would be a silly design mistake on the C++ standard committee's part if
this code is invalid, as the author of f<>() has no knowledge on where it will
be used and cannot anticipate whether it will be instantiated with a type that
has a member named T.  If the code is invalid, regardless of what name the
author of f<>() picks for its template parameter, f will fail to instantiate
for some type that happens to have a member with the same name.  This doesn't
make sense and basically makes it impossible to write truly generic functions.

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.



More information about the llvm-bugs mailing list