Bug 191057 – [Win][MSVC] Add /Zc:__cplusplus to enable __cplusplus macro standard conformance mode

• Component: Web Template Framework
• Status: RESOLVED WONTFIX

[Fujii Hironori]

[Win][MSVC] Add /Zc:__cplusplus to enable __cplusplus macro standard conformance mode

MSVC now correctly reports __cplusplus | Visual C++ Team Blog
https://blogs.msdn.microsoft.com/vcblog/2018/04/09/msvc-now-correctly-reports-__cplusplus/

[Fujii Hironori]

Created attachment 353357 [details]
Patch

[Ross Kirsling]

Comment on attachment 353357 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=353357&action=review

Informal r+ -- seems great!

Do you think there's a benefit to reducing _MSC_VER usage in a subsequent patch?

> Source/WTF/wtf/StdLibExtras.h:530
> -#if (__cplusplus < 201703L || (defined(__GLIBCXX__) && !defined(_GLIBCXX_RELEASE))) && (!defined(_MSC_FULL_VER) || _MSC_FULL_VER < 190023918)
> +#if __cplusplus < 201703L || (defined(__GLIBCXX__) && !defined(_GLIBCXX_RELEASE))

Looks like there's another instance of _MSC_FULL_VER immediately before this—shouldn't we remove it there too?

[Fujii Hironori]

(In reply to Ross Kirsling from comment #2)
> Do you think there's a benefit to reducing _MSC_VER usage in a subsequent
> patch?

To be clarify at first.
_MSC_VER and _MSC_FULL_VER are for the compiler version.
__cplusplus and _MSVC_LANG are for the language version.

_MSVC_LANG should be removed in favor of sane __cplusplus.

The original code was using _MSC_FULL_VER. This seems a simple mistake.


> > Source/WTF/wtf/StdLibExtras.h:530
> > -#if (__cplusplus < 201703L || (defined(__GLIBCXX__) && !defined(_GLIBCXX_RELEASE))) && (!defined(_MSC_FULL_VER) || _MSC_FULL_VER < 190023918)
> > +#if __cplusplus < 201703L || (defined(__GLIBCXX__) && !defined(_GLIBCXX_RELEASE))
> 
> Looks like there's another instance of _MSC_FULL_VER immediately before
> this—shouldn't we remove it there too?

Year, you are right. There is.
And, I can remove it.
But, this is not so simple.

The reason I'm doing this task is I want to downgrade Windows
port to C++14 based of webkit-dev discussion
<https://lists.webkit.org/pipermail/webkit-dev/2018-September/030186.html>.

Here are the condition.

> #if __cplusplus < 201703L && (!defined(_MSC_FULL_VER) || _MSC_FULL_VER < 190023918) && !defined(__cpp_lib_logical_traits)

I can remove the _MSC_FULL_VER check because Windows port is using C++17.
But, if I change /std:c++17 to /std:c++14. Then following compilation error happens.

> 1>------ Build started: Project: WTF (WTF\WTF), Configuration: Debug x64 ------
> 1>DecimalNumber.cpp
> 1>c:\webkit\gb\source\wtf\wtf\stdlibextras.h(525): error C2953: 'std::conjunction': class template has already been defined
> 1>c:\webkit\gb\source\wtf\wtf\stdlibextras.h(525): note: see declaration of 'std::conjunction'
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\tuple(62): error C3203: 'conjunction': unspecialized class template can't be used as a template argument for template parameter '_Trait', expected a real type
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\tuple(65): note: see reference to class template instantiation 'std::_Tuple_explicit_val0<true,std::tuple<_Types...>,_Srcs...>' being compiled
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): error C2903: 'conjunction': symbol is neither a class template nor a function template
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): note: see reference to variable template 'const bool conjunction_v<std::negation<std::is_pointer<std::default_delete<WTF::StringView::GraphemeClusters::Iterator::Impl> > >,std::is_default_constructible<std::default_delete<WTF::StringView::GraphemeClusters::Iterator::Impl> > >' being compiled
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): error C2065: 'value': undeclared identifier
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): error C2131: expression did not evaluate to a constant
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): note: failure was caused by non-constant arguments or reference to a non-constant symbol
> 1>c:\program files (x86)\microsoft visual studio\2017\professional\vc\tools\msvc\14.14.26428\include\type_traits(1364): note: see usage of 'value'
> 1>Done building project "WTF.vcxproj" -- FAILED.
> ========== Build: 0 succeeded, 1 failed, 0 up-to-date, 0 skipped ==========

std::conjunction should be defined after C++17.
This looks a MSVC bug.

[Fujii Hironori]

Oh, no. AppleWin EWS is red.

> -- The C compiler identification is MSVC 19.11.25547.0
(...)
> /cygdrive/c/Program Files (x86)/Microsoft Visual Studio/2017/Community/MSBuild/15.0/bin/MSBuild.exe /verbosity:minimal C:\cygwin\home\buildbot\WebKit\WebKitBuild\Release\WebKit.sln /t:build /p:Configuration=Release /p:Platform=Win32 /fl /flp:LogFile=C:\cygwin\home\buildbot\WebKit\WebKitBuild\Release\webkit_errors.log;ErrorsOnly /fl1 /flp1:LogFile=C:\cygwin\home\buildbot\WebKit\WebKitBuild\Release\webkit_warnings.log;WarningsOnly /maxcpucount:8
> Microsoft (R) Build Engine version 15.4.8.50001 for .NET Framework
(...)
> cl : Command line warning D9002: ignoring unknown option '/Zc:__cplusplus' [C:\cygwin\home\buildbot\WebKit\WebKitBuild\Release\Source\WTF\wtf\WTF.vcxproj]

[Fujii Hironori]

We dropped MSVC support. Closed.