CppInterOp exposes API from Clang and LLVM in a backward compatibe way. The API support downstream tools that utilize interactive C++ by using the compiler as a service. That is, embed Clang and LLVM as a libraries in their codebases. The API are designed to be minimalistic and aid non-trivial tasks such as language interoperability on the fly. In such scenarios CppInterOp can be used to provide the necessary introspection information to the other side helping the language cross talk.
The CppInterOp library provides a minimalist approachfor other languages to identify C++ entities (variables, classes, etc.). This enables interoperability with C++ code, bringing the speed and efficiency of C++ to simpler, more interactive languages like Python.
CppInterOp can be adopted incrementally. While the rest of the framework is the same, a small part of CppInterOp can be utilized. More components may be adopted over time.
While the library includes some tricky code, it is designed to be simple and robust (simple function calls, no inheritance, etc.). The goal is to make it as close to the compiler API as possible, and each routine to do just one thing that it was designed for.
The main use case for CppInterOp is with the CPPYY service. CPPYY is an automatic run-time bindings generator for Python & C++, and supports a wide range of C++ features (e.g., template instantiation). It operates on demand and generates only what is necessary. It requires a compiler (Cling1 /Clang-REPL2) that can be available during program runtime.
Once CppInterOp is integrated with LLVM's3 Clang-REPL component (that can then be used as a runtime compiler for CPPYY), it will further enhance CPPYY's performance in the following ways:
- Simpler codebase: Removal of string parsing logic will lead to a simpler code base.
- LLVM Integration: The CppInterOp interfaces will be a part of the LLVM toolchain (as part of Clang-REPL).
- Better C++ Support: C++ features such as Partial Template Specialization will be available through CppInterOp.
- Fewer Lines of Code: A lot of dependencies and workarounds will be removed, reducing the lines of code required to execute CPPYY.
- Well tested interoperability Layer: The CppInterOp interfaces have full unit test coverage.
Besides being developed as a general-purpose library, one of the long-term goals of CppInterOp is to stay backward compatible and be adopted in the High Energy Physics (HEP) field, as it will become an essential part of the Root framework. Over time, parts of the Root framework can be swapped by this API, adding speed and resilience with it.
Build instructions for CppInterOp and its dependencies are as follows.
Clone and checkout the LLVM project repository.
git clone https://github.com/llvm/llvm-project.git
git checkout release/16.x
Get the following patches required for development work.
compgen -G "../patches/llvm/clang16-*.patch" > /dev/null && find ../patches/llvm/clang16-*.patch -printf "%f\n" && git apply ../patches/llvm/clang16-*.patch
Clang-REPL is an interpreter that CppInterOp works alongside. Build Clang (and Clang-REPL along with it).
mkdir build && cd build
cmake -DLLVM_ENABLE_PROJECTS=clang \
-DLLVM_TARGETS_TO_BUILD="host;NVPTX" \
-DCMAKE_BUILD_TYPE=Release \
-DLLVM_ENABLE_ASSERTIONS=ON \
-DLLVM_USE_LINKER=lld \
-DCLANG_ENABLE_STATIC_ANALYZER=OFF \
-DCLANG_ENABLE_ARCMT=OFF \
-DCLANG_ENABLE_FORMAT=OFF \
-DCLANG_ENABLE_BOOTSTRAP=OFF \
../llvm
cmake --build . --target clang clang-repl --parallel $(nproc --all)
Note the 'llvm-project' directory location.
cd ../
export LLVM_DIR=$PWD
cd ../
Next, use the following export commands.
export CB_PYTHON_DIR="$PWD/cppyy-backend/python"
export CPPINTEROP_DIR="$CB_PYTHON_DIR/cppyy_backend"
Following steps are required to help build CppInterOp alongside the LLVM-Project.
mkdir build && cd build
export CPPINTEROP_BUILD_DIR=$PWD
cmake -DCMAKE_BUILD_TYPE=Release \
-DUSE_CLING=OFF \
-DUSE_REPL=ON \
-DLLVM_DIR=$LLVM_BUILD_DIR \
-DLLVM_USE_LINKER=lld \
-DBUILD_SHARED_LIBS=ON \
-DCMAKE_INSTALL_PREFIX=$CPPINTEROP_DIR \
../
cmake --build . --target install --parallel $(nproc --all)
Besides the Clang-REPL interpreter, CppInterOp also works alongside the Cling
interpreter. Cling depends on its own customised version of llvm-project,
hosted under the root-project (see the git path below).
Use the following build instructions.
git clone --depth=1 https://github.com/root-project/cling.git
git clone --depth=1 -b cling-llvm13 https://github.com/root-project/llvm-project.git
cd llvm-project
mkdir build && cd build
cmake -DLLVM_ENABLE_PROJECTS=clang \
-DLLVM_EXTERNAL_PROJECTS=cling \
-DLLVM_EXTERNAL_CLING_SOURCE_DIR=../../cling \
-DLLVM_TARGETS_TO_BUILD="host;NVPTX" \
-DCMAKE_BUILD_TYPE=Release \
-DLLVM_ENABLE_ASSERTIONS=ON \
-DLLVM_USE_LINKER=lld \
-DCLANG_ENABLE_STATIC_ANALYZER=OFF \
-DCLANG_ENABLE_ARCMT=OFF \
-DCLANG_ENABLE_FORMAT=OFF \
-DCLANG_ENABLE_BOOTSTRAP=OFF \
../llvm
cmake --build . --target clang --parallel $(nproc --all)
cmake --build . --target cling --parallel $(nproc --all)
cmake --build . --target gtest_main --parallel $(nproc --all)
Note the 'llvm-project' directory location.
cd ../
export LLVM_DIR=$PWD
cd ../
Next, export the following directory.
export CPPINTEROP_DIR=$PWD/cppyy-backend/python/cppyy_backend
Finally, clone the CppInterOp repository.
git clone https://github.com/compiler-research/CppInterOp.git
cd CppInterOp
mkdir build && cd build
CPPINTEROP_BUILD_DIR=$(PWD)
Execute the following.
cmake -DBUILD_SHARED_LIBS=ON -DUSE_CLING=ON -DUSE_REPL=Off -DCling_DIR=$LLVM_DIR/build -DCMAKE_INSTALL_PREFIX=$CPPINTEROP_DIR ..
cmake --build . --target install
Further Reading: C++ Language Interoperability Layer