Boost C++ Libraries: Ticket #11338: mingw warnings

component changed; owner set

John Maddock — Thu, 28 May 2015 16:13:02 GMT

owner set to Andrey Semashev
component None → winapi

owner, component changed

Andrey Semashev — Mon, 01 Jun 2015 07:51:22 GMT

owner changed from Andrey Semashev to viboes
component winapi → thread

IIRC, Boost.Thread developers were planning to port to the Boost.WinAPI declarations at some point. This would remove the discrepancies between the declarations and, I think, this would be the right solution.

cc set

Andrey Semashev — Mon, 01 Jun 2015 09:18:34 GMT

cc Andrey.Semashev@… added

viboes — Mon, 01 Jun 2015 17:10:20 GMT

I have no access to a windows environment.

I could create a branch and do my best to port to Boost.WinAPI if someone is ready to test it.

I would also accept any working patch.

Edward Diener — Mon, 01 Jun 2015 21:17:59 GMT

I can test the changes made to Boost.WinAPI on Windows 7 and/or 8.1 using VC++ ( 8 through 12 ), gcc ( 4.3 through 4.9.2 ), and the latest clang. Just tell me what the git branch is and I will test it for you when you are finished with it.

viboes — Mon, 01 Jun 2015 21:26:38 GMT

Thanks, I will post it here.

Edward Diener — Tue, 02 Jun 2015 11:28:26 GMT

My clang bug report at https://llvm.org/bugs/show_bug.cgi?id=23722 came back with:

The problem is that the two function declarations have different parameter types. The windows.h version takes ::_FILETIME*, whereas the boost version takes boost::detail::winapi::_FILETIME*. Because they're 'extern "C"' functions, they are redeclarations of the same function, and are ill-formed because they have different types.

It looks like the only viable solution which will satisfy clang, and the C++ standard, is along the lines of what Peter Dimov suggested for all the winapi functionality:

/* at global scope */
  struct FILETIME;
  extern "C" __declspec(dllimport) void __stdcall GetSystemTimeAsFileTime(FILETIME*);
  namespace winapi
  {
     struct FILETIME { ... };
     inline void GetSystemTimeAsFileTime( FILETIME * p )
     {
         ::GetSystemTimeAsFileTime( (::FILETIME*)p );
     }
  }

and then calling:

  boost::winapi::FILETIME ft;
  boost::winapi::GetSystemTimeAsFileTime(&ft);

is no problem for clang and the C++ standard.

Andrey Semashev — Tue, 02 Jun 2015 11:46:56 GMT

We'll wait and see if the clang behavior is actually backed by the standard.

I think Boost.Thread needs to be ported to Boost.WinAPI regardless of the clang issue, at least to remove duplication. We can deal with the clang issue in Boost.WinAPI, and once Boost.Thread is ported, the problem will be solved for it as well.

Edward Diener — Fri, 05 Jun 2015 09:47:25 GMT

The clang explanation is:

[dcl.link]/6: "At most one function with a particular name can have C language linkage. Two declarations for a function with C language linkage with the same function name (ignoring the namespace names that qualify it) that appear in different namespace scopes refer to the same function."

[basic.link]/10: "the types specified by all declarations referring to a given variable or function shall be identical"

So the program is ill-formed because the same function is declared with two different types.

status changed

viboes — Sun, 28 Jun 2015 09:10:36 GMT

status new → assigned

sonoro@… — Tue, 16 Feb 2016 15:07:59 GMT

Replying to viboes: Is this still alive?

status changed; resolution set

viboes — Tue, 14 Nov 2017 20:53:45 GMT

status assigned → closed
resolution → obsolete

Closed as Boost.Thread uses now Boost.WinApi