# \[Solved?\] Building 7.3 on Windows **URL:** \ **Category:** FoundationDB Core\ **Created:** [October 3, 2023, 7:58pm UTC](https://forums.foundationdb.org/t/solved-building-7-3-on-windows/4176 "2023-10-03T19:58:25Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![Imperatorx](https://avatars.discourse-cdn.com/v4/letter/i/d2c977/32.png) [@Imperatorx](https://forums.foundationdb.org/u/Imperatorx)\ **Post date:** [October 3, 2023, 7:58pm UTC](https://forums.foundationdb.org/t/solved-building-7-3-on-windows/4176/1 "2023-10-03T19:58:25Z")
As the release of 7.3 is closing in, I’m trying to build it on windows for development purposes (and to create the fdb\_java.dll for JNI). It was working fine for 7.1 (without even building boost), but with 7.3 I get strange static linking errors: I’m using VS2019 as the readme suggests, but I had to upgrade boost to 1.78 for FDB 7.3. Building boost: ```auto cd c:\boost .\bootstrap.bat --with-libraries=all .\b2 link=static cxxflags="-std=c++14 -stdlib=libc++ -nostdlib++ -fPIC" linkflags="-stdlib=libc++ -nostdlib++ -static-libgcc -lc++ -lc++abi" -j32 install ``` Building FDB: ```auto cd c:\build cmake -G "Visual Studio 16 2019" -A x64 -T ClangCl c:\foundationdb msbuild /p:Configuration=Release foundationdb.sln /p:UseMultiToolTask=true /p:CL_MPCount=8 ``` Error: ```auto lld-link : error : /failifmismatch: mismatch detected for 'RuntimeLibrary': [c:\build\bindings\c\fdb_c.vcxproj] >>> flow.lib(JsonTraceLogFormatter.obj) has value MT_StaticRelease >>> libboost_filesystem-vc142-mt-x64-1_78.lib(operations.obj) has value MD_DynamicRelease ``` This seems strange, as I have built boost to be static (mt in the filename also suggests that), but cmake is complaining that is dyamic? Has anyone succeeded to build it on windows? ---
**Author:** ![Imperatorx](https://avatars.discourse-cdn.com/v4/letter/i/d2c977/32.png) [@Imperatorx](https://forums.foundationdb.org/u/Imperatorx)\ **Post date:** [October 4, 2023, 2:53pm UTC](https://forums.foundationdb.org/t/solved-building-7-3-on-windows/4176/2 "2023-10-04T14:53:01Z")
I managed to progress forward with: ```auto Install Python 3.12 Install CMake: 3.27.6 Install Boost: 1.78 python3 -m pip install setuptools==65.3.0 sphinx-bootstrap-theme==0.8.1 docutils==0.19 sphinx==5.1.1 sphinx-autobuild Jinja2==3.1.2 urllib3==2.0.2 Building boost: .\bootstrap.bat --with-libraries=all .\b2 cxxflags="-std=c++14 -stdlib=libc++ -nostdlib++ -fPIC" linkflags="-stdlib=libc++ -nostdlib++ -static-libgcc -static -lc++ -lc++abi" -j32 link=static threading=multi runtime-link=static release install Building FDB: cd c:\ mkdir c:\build git clone --depth 1 --branch 7.3.19 https://github.com/apple/foundationdb.git cd c:\build && cmake -G "Visual Studio 16 2019" -A x64 -T ClangCl c:\foundationdb msbuild /p:Configuration=Release foundationdb.sln /p:UseMultiToolTask=true /p:CL_MPCount=4 ``` After this, the `fdb_c.dll`, the `fdb_java.dll` and the `fdbcli` binaries are correctly built, but the build of the `fdbserver` fails with: ```auto In file included from C:/foundationdb/fdbserver/workloads/workloads.actor.cpp:25: In file included from C:\foundationdb\fdbclient\include\fdbclient/DataDistributionConfig.actor.h:23: In file included from C:/foundationdb/fdbclient/include/fdbclient/DataDistributionConfig.actor.h:32: In file included from C:\foundationdb\fdbclient\include\fdbclient/KeyBackedRangeMap.actor.h:24: C:/foundationdb/fdbclient/include/fdbclient/KeyBackedRangeMap.actor.h(131,100): error : no matching constructor for initialization of 'ObjectCodec' [c:\build\fdbserver\fdbserver.vcxproj] C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.29.30133\include\type_traits(598,49): note: in instantiation of default function argument expression for 'KeyBackedRangeMap, DDRangeConfig, TupleCodec>, ObjectCodec>' required here C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.29.30133\include\optional(170,44): note: in instantiation of variable template specialization 'std::is_constructible_v, DDRangeConfig, TupleCodec>, ObjectCodec>>' requested here C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.29.30133\include\optional(171,24): note: while substituting prior template arguments into non-type template parameter [with _Types = <>] C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.29.30133\include\optional(155,7): note: while substituting deduced template arguments into function template 'optional' [with _Types = <>, $1 = (no ``` Any ideas? ---
**Author:** ![Imperatorx](https://avatars.discourse-cdn.com/v4/letter/i/d2c977/32.png) [@Imperatorx](https://forums.foundationdb.org/u/Imperatorx)\ **Post date:** [October 19, 2023, 3:38pm UTC](https://forums.foundationdb.org/t/solved-building-7-3-on-windows/4176/3 "2023-10-19T15:38:22Z")
I managed to build it (windows installer, JNI DLL) after modifying the source to create default initializers for the following structs: - for `VersionOptions` in `ObjectCodec` - for `ProtocolVersion` in `IncludeVersion`’ These are used in the `KeyBackedRangeMap`. Example: ```auto --- a/fdbclient/include/fdbclient/KeyBackedTypes.actor.h +++ b/fdbclient/include/fdbclient/KeyBackedTypes.actor.h @@ -186,7 +186,7 @@ struct BinaryCodec { // Codec for using Flatbuffer compatible types via ObjectWriter/ObjectReader template struct ObjectCodec { - ObjectCodec(VersionOptions vo) : vo(vo) {} + ObjectCodec(VersionOptions vo = VersionOptions{}) : vo(vo) {} VersionOptions vo; inline Standalone pack(T const& val) const { return ObjectWriter::toValue(val, vo); } ``` I built it using VS2022 instead of VS2019, so it complained about a bunch of `==` comparisons between a structs and Optional-s: in - tester.actor.cpp - DcLag.actor.cpp - GcGenerations.actor.cpp I don’t know c++, but are these comparisons handled on linux via operator overloading, or are these returning always false (a bug)? Example: ```auto --- a/fdbserver/workloads/GcGenerations.actor.cpp +++ b/fdbserver/workloads/GcGenerations.actor.cpp @@ -97,7 +97,7 @@ struct GcGenerationsWorkload : TestWorkload { std::vector remoteIps; // all remote process IPs for (const auto& process : g_simulator->getAllProcesses()) { const auto& ip = process->address.ip; - if (process->locality.dcId().present() && process->locality.dcId().get() == g_simulator->remoteDcId && + if (process->locality.dcId().present() && process->locality.dcId().get() == g_simulator->remoteDcId.get() && !isCoordinator(coordinators, ip)) { remoteIps.push_back(ip); } else { diff --git a/flow/include/flow/serialize.h b/flow/include/flow/serialize.h ``` ---
**Author:** ![jzhou](https://sea1.discourse-cdn.com/foundationdb/user_avatar/forums.foundationdb.org/jzhou/32/445_2.png) [@jzhou](https://forums.foundationdb.org/u/jzhou)\ **Post date:** [October 21, 2023, 4:49am UTC](https://forums.foundationdb.org/t/solved-building-7-3-on-windows/4176/4 "2023-10-21T04:49:46Z")
The latest `main` branch should have fixed these issues, mainly in [Fixed compiling foundationdb with the clang 17 compiler by oleg68 · Pull Request #10518 · apple/foundationdb · GitHub](https://github.com/apple/foundationdb/pull/10518)