There are plenty of existing libraries that excel at very specific tasks. It's generally a very good idea to reuse such libraries in our own codebases, because we can rely on years of experience from other groups of experts. As computer architectures and compilers evolve, so do programming languages. Whereas years ago most scientific software was written in Fortran, nowadays C, C++, and interpreted languages – first and foremost Python – are taking the center stage. It is indeed more and more common to integrate code written in a compiled language with bindings to an interpreted language, since it affords the following benefits:

• End-users can customize and expand the capabilities offered by the code itself to fully suit their needs.
• One is able to combine the expressiveness of a language such as Python with the performance of a compiled language that is closer "to the metal" in terms of memory addressing, getting the best of both worlds.

As we have consistently shown throughout the previous recipes, the project command can be used to set the languages used in the project via the LANGUAGES keyword. CMake has support for many – but not all – compiled programming languages. As of CMake 3.5, various flavors of assembly (such as ASM-ATT, ASM, ASM-MASM, and ASM-NASM), C, C++, Fortran, Java, RC (Windows Resource Compiler), and Swift are valid choices. CMake 3.8 added support for two more languages: C# and CUDA (see the release notes here: https://cmake.org/cmake/help/v3.8/release/3.8.html#languages).

In this chapter, we will show how to integrate codes written in different compiled (C, C++, and Fortran) and interpreted (Python) languages in a way that is portable and cross-platform. We will show how to leverage CMake and tools intrinsic to the different programming languages we aim to integrate.