[libcxx] r292294 - Fix type_info's constructor by making it explicit again.

[Eric Fiselier via cfe-commits]

Merged into the 4.0 branch in r292309.

On Tue, Jan 17, 2017 at 4:55 PM, Eric Fiselier <eric at efcs.ca> wrote:

> FYI Once the bots finish checking this commit I'm going to merge this into
> the 4.0 branch.
>
> This fixes a very recently introduced regression.
>
> /Eric
>
> On Tue, Jan 17, 2017 at 4:41 PM, Eric Fiselier via cfe-commits <
> cfe-commits at lists.llvm.org> wrote:
>
>> Author: ericwf
>> Date: Tue Jan 17 17:41:42 2017
>> New Revision: 292294
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=292294&view=rev
>> Log:
>> Fix type_info's constructor by making it explicit again.
>>
>> In recent changes type_info's private constructor was
>> accidentally made implicit. This patch fixes that.
>>
>> Modified:
>>     libcxx/trunk/include/typeinfo
>>     libcxx/trunk/test/std/language.support/support.rtti/type.
>> info/type_info.pass.cpp
>>
>> Modified: libcxx/trunk/include/typeinfo
>> URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/include/typ
>> einfo?rev=292294&r1=292293&r2=292294&view=diff
>> ============================================================
>> ==================
>> --- libcxx/trunk/include/typeinfo (original)
>> +++ libcxx/trunk/include/typeinfo Tue Jan 17 17:41:42 2017
>> @@ -95,12 +95,13 @@ protected:
>>      uintptr_t __type_name;
>>
>>      _LIBCPP_INLINE_VISIBILITY
>> -    type_info(const char* __n) : __type_name(reinterpret_cast<uintptr_t>(__n))
>> {}
>> +    explicit type_info(const char* __n)
>> +      : __type_name(reinterpret_cast<uintptr_t>(__n)) {}
>>  #else
>>      const char *__type_name;
>>
>>      _LIBCPP_INLINE_VISIBILITY
>> -    type_info(const char* __n) : __type_name(__n) {}
>> +    explicit type_info(const char* __n) : __type_name(__n) {}
>>  #endif
>>
>>  public:
>>
>> Modified: libcxx/trunk/test/std/language.support/support.rtti/type.
>> info/type_info.pass.cpp
>> URL: http://llvm.org/viewvc/llvm-project/libcxx/trunk/test/std/la
>> nguage.support/support.rtti/type.info/type_info.pass.cpp?rev
>> =292294&r1=292293&r2=292294&view=diff
>> ============================================================
>> ==================
>> --- libcxx/trunk/test/std/language.support/support.rtti/type.
>> info/type_info.pass.cpp (original)
>> +++ libcxx/trunk/test/std/language.support/support.rtti/type.
>> info/type_info.pass.cpp Tue Jan 17 17:41:42 2017
>> @@ -10,11 +10,16 @@
>>  // test type_info
>>
>>  #include <typeinfo>
>> +#include <string>
>>  #include <cstring>
>>  #include <cassert>
>>
>> +bool test_constructor_explicit(std::type_info const&) { return false; }
>> +bool test_constructor_explicit(std::string const&) { return true; }
>> +
>>  int main()
>>  {
>> +  {
>>      const std::type_info& t1 = typeid(int);
>>      const std::type_info& t2 = typeid(int);
>>      assert(t1 == t2);
>> @@ -23,4 +28,13 @@ int main()
>>      assert(!t1.before(t2));
>>      assert(strcmp(t1.name(), t2.name()) == 0);
>>      assert(strcmp(t1.name(), t3.name()) != 0);
>> +  }
>> +  {
>> +    // type_info has a protected constructor taking a string literal.
>> This
>> +    // constructor is not intended for users. However it still
>> participates
>> +    // in overload resolution, so we need to ensure that it is marked
>> explicit
>> +    // to avoid ambiguous conversions.
>> +    // See: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216201
>> +    assert(test_constructor_explicit("abc"));
>> +  }
>>  }
>>
>>
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20170117/4bd7cf50/attachment.html>