Bug 173889

Created attachment 313946 [details] Patch to build jsc-only with clang-cl toolset

Created attachment 313952 [details] Patch to build jsc-only with clang-cl toolset

Created attachment 313956 [details] Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 option

Comment on attachment 313956 [details] Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 option View in context: https://bugs.webkit.org/attachment.cgi?id=313956&action=review > Source/cmake/OptionsCommon.cmake:57 > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions -fno-strict-aliasing -fno-rtti") Don't duplicate these lines > Source/cmake/OptionsCommon.cmake:58 > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++1y") Can clang-cl accept -std=c++1y? If yes, it should just add a few flags to those set in the common COMPILER_IS_GCC_OR_CLANG block

Comment on attachment 313956 [details] Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 option View in context: https://bugs.webkit.org/attachment.cgi?id=313956&action=review >> Source/cmake/OptionsCommon.cmake:57 >> + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions -fno-strict-aliasing -fno-rtti") > > Don't duplicate these lines These were duplicated from above. I'm guessing because those options are never going to change. >> Source/cmake/OptionsCommon.cmake:58 >> + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++1y") > > Can clang-cl accept -std=c++1y? If yes, it should just add a few flags to those set in the common COMPILER_IS_GCC_OR_CLANG block With clang-cl it attempts to use the cl options where applicable such as https://blogs.msdn.microsoft.com/vcblog/2016/06/07/standards-version-switches-in-the-compiler/

(In reply to Konstantin Tokarev from comment #4) > Comment on attachment 313956 [details] > Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 > option > > View in context: > https://bugs.webkit.org/attachment.cgi?id=313956&action=review > > > Source/cmake/OptionsCommon.cmake:57 > > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions -fno-strict-aliasing -fno-rtti") > > Don't duplicate these lines > > Source/cmake/OptionsCommon.cmake:58 > > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++1y") > > Can clang-cl accept -std=c++1y? If yes, it should just add a few flags to > those set in the common COMPILER_IS_GCC_OR_CLANG block Unfortunately, no, -std is considered an unknown argument. (In reply to Konstantin Tokarev from comment #4) > Comment on attachment 313956 [details] > Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 > option > > View in context: > https://bugs.webkit.org/attachment.cgi?id=313956&action=review > > > Source/cmake/OptionsCommon.cmake:57 > > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions -fno-strict-aliasing -fno-rtti") > > Don't duplicate these lines I'm sorry, I'm not sure what you mean here. Is it because we're doing -fno-strict-aliasing in both branches? We aren't passing -fno-exceptions or -fno-rtti in the clang-cl case, but it'd be easy to separate the common one out if that seems better. > > Source/cmake/OptionsCommon.cmake:58 > > + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++1y") > > Can clang-cl accept -std=c++1y? If yes, it should just add a few flags to > those set in the common COMPILER_IS_GCC_OR_CLANG block Unfortunately, no, -std is an unknown argument for clang-cl.

Sorry, I accidentally ended up with two copies of the last post in the post.

Comment on attachment 313956 [details] Patch to build jsc-only with clang-cl toolset - combined clang-cl/gcc/clang4 option View in context: https://bugs.webkit.org/attachment.cgi?id=313956&action=review >>>>> Source/cmake/OptionsCommon.cmake:57 >>>>> + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions -fno-strict-aliasing -fno-rtti") >>>> >>>> Don't duplicate these lines >>> >>> These were duplicated from above. I'm guessing because those options are never going to change. >> >> > > I'm sorry, I'm not sure what you mean here. Is it because we're doing -fno-strict-aliasing in both branches? We aren't passing -fno-exceptions or -fno-rtti in the clang-cl case, but it'd be easy to separate the common one out if that seems better. Yes, it would be better

Created attachment 314059 [details] Patch to build jsc-only with clang-cl toolset - restructured gcc_or_clang block

Created attachment 314060 [details] Patch to build jsc-only with clang-cl toolset - restructured gcc_or_clang block

Comment on attachment 314060 [details] Patch to build jsc-only with clang-cl toolset - restructured gcc_or_clang block Clearing flags on attachment: 314060 Committed r218900: <http://trac.webkit.org/changeset/218900>

All reviewed patches have been landed. Closing bug.