Re: Strongly typed enums as bit fields

Re: Strongly typed enums as bit fields

[Владимир Андреев]

Hello all!

I try to compile the following test program using GCC 4.6.1:

enum class MyEnum: unsigned char
{
  A = 1,
  B = 2
};

struct MyClass
{
  MyEnum Field1: 3;
  unsigned char Field2: 5;
};

int main()
{
  MyClass Class;
}

but have error:

C:\MinGW\bin>gcc E:/Test/Main.cpp -o E:/Test/Out.exe -std=c++0x
E:/Test/Main.cpp:9:18: error: bit-field 'Field1' with non-integral type

MSVC 11.0 normally processes this program.

Why GCC throws this error and MCVC doesn't?

--
With regards, Vladimir

Re: Strongly typed enums as bit fields

[Jonathan Wakely]

2011/11/20 Владимир Андреев:
>
> C:\MinGW\bin>gcc E:/Test/Main.cpp -o E:/Test/Out.exe -std=c++0x
> E:/Test/Main.cpp:9:18: error: bit-field 'Field1' with non-integral type
>
> MSVC 11.0 normally processes this program.
>
> Why GCC throws this error and MCVC doesn't?

I think this should be valid, the standard says "A bit-field shall have
integral or enumeration type (3.9.1)."

Please report it to bugzilla, thanks.

Re: Strongly typed enums as bit fields

[Владимир Андреев]

Today I have tested Intel C++ Compiler - that is OK.

Additionally I noted, that GCC doesn't support declaration like

using NewType = OldType;

which semantically equals to

typedef OldType NewType;

2011/11/20 Jonathan Wakely <jwakely.gcc@gmail.com>:
> 2011/11/20 Владимир Андреев:
>>
>> C:\MinGW\bin>gcc E:/Test/Main.cpp -o E:/Test/Out.exe -std=c++0x
>> E:/Test/Main.cpp:9:18: error: bit-field 'Field1' with non-integral type
>>
>> MSVC 11.0 normally processes this program.
>>
>> Why GCC throws this error and MCVC doesn't?
>
> I think this should be valid, the standard says "A bit-field shall have
> integral or enumeration type (3.9.1)."
>
> Please report it to bugzilla, thanks.
>



-- 
С уважением, Владимир

Re: Strongly typed enums as bit fields

[Jonathan Wakely]

2011/11/20 Владимир Андреев :
> Additionally I noted, that GCC doesn't support declaration like
>
> using NewType = OldType;
>
> which semantically equals to
>
> typedef OldType NewType;

Yes it does, but that was only added last week.

Thanks for the Bugzilla report.