Compiler errors when specifying a target Java source file

[rafeememon]

I'm working on a JavaCPP wrapper around the nanodbc library for connecting to ODBC data sources from Java. I'm bringing the library in via a git submodule and using gradle-javacpp to build the wrapper. I'm using Windows 11 and Visual Studio 2022 Community. I'm running into a couple issues, but I'm not exactly sure how/if they are related.

Repo: https://github.com/rafeememon/nanodbc-java-test/tree/f10465816f3fdae03553834d0d26353063a9e2db

I have a basic mapping configured in NanodbcConfig.java. In its current state, ./gradlew build completes successfully, and jninanodbc.dll appears in the expected platform subdirectory under build/classes. However, no wrapper .java file is generated to interface with the native library.

Adding a target in the @Properties annotation of the mapping by uncommenting this line results in the wrapper .java file being generated, but I also encounter a slew of new compiler errors in the generated jninanodbc.cpp file. It seems as though the errors are for illegal member accesses that should have been #defined away by my compiler options added to the build tasks here.

Build log without specifying target: gradle_build_without_target.log
Build log with specifying target: gradle_build_with_target.log

Is there something I should be doing differently to properly generated the wrapper .java file? Thanks for any guidance you can provide!

[saudet]

We need to tell JavaCPP not to try to translate to variables any of those #define macros manually:
https://github.com/bytedeco/javacpp/wiki/Mapping-Recipes#ignoring-attributes-and-macros
It's a bit tedious, but once it's done it doesn't usually take a lot of time to maintain.

[saudet]

If that's a C API, you could use JNA, JNR, Panama, etc.

[rafeememon]

Thanks for the pointers. I've made a lot of progress in getting my project to compile, but I'm running into another issue that's stumping me. It seems that for the functions dealing with strings, either as parameters or returned values, the compiler isn't able to properly construct a StringAdapter. I suspect I need to add a .cast() somewhere, but I'm not exactly sure.

Console output from javacppBuildCompiler: https://scans.gradle.com/s/ymjjizojn7ciu/console-log?task=:javacppBuildCompiler
Updated repo: https://github.com/rafeememon/nanodbc-java-test/tree/c9d591caf9cc545b659585f2310a2cd9ce70cf21

D:\git\nanodbc-java-test\build\classes\java\main\io\nanodbc\jniNanodbcNative.cpp(499): error C2446: ':': no conversion from 'const wchar_t [1]' to 'T *'
        with
        [
            T=char32_t
        ]
...
D:\git\nanodbc-java-test\build\classes\java\main\io\nanodbc\jniNanodbcNative.cpp(499): error C2665: 'std::basic_string<char32_t,std::char_traits<char32_t>,std::allocator<char32_t>>::basic_string': no overloaded function could convert all the argument types

The errors seem to come from the following constructor:

StringAdapter(const   signed   int* ptr, typename std::basic_string<T>::size_type size, void* owner) : ptr((T*)ptr), size(size), owner(owner),
        str2(ptr ? (T*)ptr : L"", ptr ? (size > 0 ? size : wcslen((wchar_t*)ptr)) : 0), str(str2) { }

[saudet]

Right, that's a tricky one. We need to get a pointer out of that.

[saudet]

I'm not seeing any "wchar_t" inside the header file. Where is it coming from?

[rafeememon]

That's a good hint... It looks like it was coming about because nanodbc::string is typedef'ed to std::wstring. The mapping I needed to make the compiler happy was:

.put(new Info("nanodbc::string").cppTypes("std::wstring").annotations("@StdWString"))

Thanks for pointing me in the right direction!