Boost C++ Libraries: Ticket #5127: Incorrect result_of usage in transform_iterator

attachment set

mimomorin@… — Thu, 27 Jan 2011 01:03:17 GMT

attachment → transform_iterator.patch

A patch for transform_iterator.hpp (against trunk)

attachment set

mimomorin@… — Thu, 27 Jan 2011 01:03:58 GMT

attachment → transform_iterator_test.patch

A patch for transform_iterator_test.cpp (against trunk)

attachment set

mimomorin@… — Thu, 27 Jan 2011 01:04:51 GMT

attachment → transform_iterator_doc.patch

A patch for transform_iterator_test's documents (against trunk)

mimomorin@… — Thu, 27 Jan 2011 03:49:06 GMT

The documentation patch (transform_iterator_doc.patch) fixes only .html and .rst files, and it doesn't touch .pdf file. Should libs/iterator/doc/transform_iterator.pdf be regenerated?

milestone changed

mimomorin@… — Thu, 27 Jan 2011 13:59:31 GMT

milestone To Be Determined → Boost 1.47.0

summary changed

mimomorin@… — Wed, 02 Feb 2011 05:27:55 GMT

summary Documentation of transform_iterator is incorrect → Incorrect result_of usage in transform_iterator

Fri, 09 Sep 2011 10:58:11 GMT

Ping?

The test runs successfully on gcc 4.2 - 4.6 and clang (trunk).

jeffrey.hellrung — Sun, 22 Apr 2012 01:27:52 GMT

(In [78121]) refs #5127 applying Michael Morin's patch for transform_iterator to trunk

status changed; resolution set

jeffrey.hellrung — Tue, 24 Apr 2012 21:28:14 GMT

status new → closed
resolution → fixed

(In [78184]) merging from trunk; fix #5127 from M. Morin; fix for refs #5697

TLBettens@… — Sat, 18 Aug 2012 21:16:47 GMT

Warning message introduced into release 1_50_0

This change went into release 1-50_0 and now causes a warning message to be emitted in Visual C++ version 10, perhaps others versions as well.

Compiler warning message:

2>c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(49): warning C4180: qualifier applied to function type has no meaning; ignored
2>          c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(70) : see reference to class template instantiation 'boost::detail::transform_iterator_base<UnaryFunc,Iterator,Reference,Value>' being compiled

Workaround is to either revert back to release 1_49_0 by removing the const:

boost::result_of</*const*/ UnaryFunction(Iterator::reference)>

, or disable the warning message using something like:

        #if defined(BOOST_MSVC)
        #   pragma warning (push)
        #   pragma warning (disable:4180) // warning C4180: qualifier applied to function type has no meaning; ignored
        #endif
          typedef typename ia_dflt_help<
              Reference
            , result_of<const UnaryFunc(typename std::iterator_traits<Iterator>::reference)>
          >::type reference;
        #if defined(BOOST_MSVC)
        #   pragma warning (pop)
        #endif

Dave Abrahams — Sat, 18 Aug 2012 23:56:13 GMT

I support the latter change

Michel Morin — Tue, 21 Aug 2012 14:57:20 GMT

Replying to TLBettens@…:

This change went into release 1-50_0 and now causes a warning message to be emitted in Visual C++ version 10, perhaps others versions as well.

Compiler warning message:
c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(49):
warning C4180: qualifier applied to function type has no meaning; ignored

Thanks for the report.

Hmm, it's strange... IIUC, UnaryFunc cannot be a function type. Could you post a (minimal) reproducible code for this problem?

anonymous — Thu, 23 Aug 2012 02:29:30 GMT

Replying to michel:

Replying to TLBettens@…:
This change went into release 1-50_0 and now causes a warning message to be emitted in Visual C++ version 10, perhaps others versions as well.

Compiler warning message:
c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(49):
warning C4180: qualifier applied to function type has no meaning; ignored
Thanks for the report.

Hmm, it's strange... IIUC, UnaryFunc cannot be a function type. Could you post a (minimal) reproducible code for this problem?

Thanks Michel for looking into it. I was able to reproduce the warning message with this code in main.cpp:

#include <boost/algorithm/string.hpp>
int main()
{
  return 0;
}

Although not too exciting but perhaps helpful, the build log (CPPRocksProject.log) capturing the compile switches and warning message looks like:

Build started 8/22/2012 7:04:55 PM.
     1>Project "D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\CPPRocksProject.vcxproj" on node 2 (build target(s)).
     1>InitializeBuildStatus:
         Creating "Win32\Debug\SAGR.unsuccessfulbuild" because "AlwaysCreate" was specified.
       MakeDirsForCl:
         Creating directory "D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\dumy\dumy\..\..\SourceCode".
       ClCompile:
         C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\CL.exe /c /I"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\..\SourceCode\\" /IC:\BoostBuilding\boost_1_50_0\ /ZI /nologo /W4 /WX- /Od /Oy- /D WIN32 /D _DEBUG /D _CONSOLE /D _MBCS /Gm /EHsc /RTCc /RTC1 /MTd /GS /Gy /fp:precise /fp:except /Za /Zc:wchar_t /Zc:forScope /GR /Fo"Win32\Debug\dumy\dumy\..\..\SourceCode\\" /Fd"Win32\Debug\vc100.pdb" /FR"Win32\Debug\dumy\dumy\..\..\SourceCode\\" /Gd /TP /wd4290 /wd4224 /analyze- /errorReport:prompt ..\..\SourceCode\main.cpp
         main.cpp
     1>c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(49): warning C4180: qualifier applied to function type has no meaning; ignored
                 c:\boostbuilding\boost_1_50_0\boost\iterator\transform_iterator.hpp(70) : see reference to class template instantiation 'boost::detail::transform_iterator_base<UnaryFunc,Iterator,Reference,Value>' being compiled
       ManifestResourceCompile:
         c:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin\rc.exe /nologo /fo"Win32\Debug\SAGR-x86-Debug.exe.embed.manifest.res" Win32\Debug\SAGR-x86-Debug_manifest.rc
       Link:
         C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\link.exe /ERRORREPORT:PROMPT /OUT:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.exe" /INCREMENTAL /NOLOGO /LIBPATH:C:\BoostBuilding\boost_1_50_0\lib32\ kernel32.lib user32.lib gdi32.lib winspool.lib comdlg32.lib advapi32.lib shell32.lib ole32.lib oleaut32.lib uuid.lib odbc32.lib odbccp32.lib /MANIFEST /ManifestFile:"Win32\Debug\SAGR-x86-Debug.exe.intermediate.manifest" /MANIFESTUAC:"level='asInvoker' uiAccess='false'" /DEBUG /PDB:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.pdb" /SUBSYSTEM:CONSOLE /TLBID:1 /DYNAMICBASE /NXCOMPAT /IMPLIB:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.lib" /MACHINE:X86 "Win32\Debug\SAGR-x86-Debug.exe.embed.manifest.res"
         Win32\Debug\dumy\dumy\..\..\SourceCode\main.obj
       Manifest:
         c:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin\mt.exe /nologo /verbose /out:"Win32\Debug\SAGR-x86-Debug.exe.embed.manifest" /manifest "Win32\Debug\SAGR-x86-Debug.exe.intermediate.manifest"
         c:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin\rc.exe /nologo /fo"Win32\Debug\SAGR-x86-Debug.exe.embed.manifest.res" Win32\Debug\SAGR-x86-Debug_manifest.rc
       LinkEmbedManifest:
         C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\link.exe /ERRORREPORT:PROMPT /OUT:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.exe" /INCREMENTAL /NOLOGO /LIBPATH:C:\BoostBuilding\boost_1_50_0\lib32\ kernel32.lib user32.lib gdi32.lib winspool.lib comdlg32.lib advapi32.lib shell32.lib ole32.lib oleaut32.lib uuid.lib odbc32.lib odbccp32.lib /MANIFEST /ManifestFile:"Win32\Debug\SAGR-x86-Debug.exe.intermediate.manifest" /MANIFESTUAC:"level='asInvoker' uiAccess='false'" /DEBUG /PDB:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.pdb" /SUBSYSTEM:CONSOLE /TLBID:1 /DYNAMICBASE /NXCOMPAT /IMPLIB:"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.lib" /MACHINE:X86 "Win32\Debug\SAGR-x86-Debug.exe.embed.manifest.res"
         Win32\Debug\dumy\dumy\..\..\SourceCode\main.obj
         CPPRocksProject.vcxproj -> D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.exe
       BscMake:
         C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\bscmake.exe /nologo /o"D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\Win32\Debug\SAGR-x86-Debug.bsc" Win32\Debug\dumy\dumy\..\..\SourceCode\main.sbr
       FinalizeBuildStatus:
         Deleting file "Win32\Debug\SAGR.unsuccessfulbuild".
         Touching "Win32\Debug\SAGR.lastbuildstate".
     1>Done Building Project "D:\SAGR RIC v3\SAGR Development\MSVC++ Solution Space (2010)\CPPRocksProject\CPPRocksProject.vcxproj" (build target(s)).
Build succeeded.
Time Elapsed 00:00:06.67

Michel Morin — Fri, 24 Aug 2012 14:52:23 GMT

OK, I figured out what the problem is. The problem is caused by the /Za compiler option. The /Za option is buggy and its use is not recommended (or rather, you shouldn't use it). By removing /Za, the warning gets disappeared. There is nothing wrong with the code in transform_iterator.hpp, and I think this warning problem would be "won't fix".

The solution to this problem is

Don't use /Za.

See the posts from Stephan T. Lavavej (Visual C++ libraries developer) such as this and this. Also, Boost's warning guidelines might be useful for you.

TLBettens@… — Fri, 24 Aug 2012 15:04:54 GMT

Thanks. but I have to disagree. Turning off strict ISO/IEC compliance, i.e. enabling Microsoft extensions beyond the standard, isn't an advisable option for portable code. Let's not rely on a particular vender's extensions. We, as a community, really should be using any help from a compiler that helps us understand when we drift away. It is what it is, and if it's unfix able okay, but really, disabling the warning message seems prudent. Either way, thanks for considering it.

TLBettens@… — Fri, 24 Aug 2012 15:15:02 GMT

That said, if the /Za option is broken in VC 2010, it hard to argue with trying to use it. Thanks again!! Michel, I'll disable the extension.