Conclusion from OpenGL Example

OpenGL has a huge API, and we integrated it into Py++ with hardly any code (mostly with configuration files). Even though the configuration files were long in some cases, they were fairly straightforward for someone familiar with the transpiler config feature. The code we did write was also very straightforward.

The challenging part was not writing the code or the configuration files; it was designing the API of our library. Once you decide on a design, that design will dictate exactly what you put in your configuration files and what small Python and C++ things you have to implement. The design was challenging because you have to be quite familiar with the following to make a good design:

• What code is possible to write in Py++
• The Py++ transpiler config feature
• The API of the C++ library you are integrating.

Also, we should note that for our example library, there might be some OpenGL functions that don't work with it. But what we did is a good start, and we can add the functions that don't work later when they are identified. Additionally, most of the functions, especially the common ones, do work.

Considering these things, I conclude that integrating C++ libraries, or parts of them, into the Py++ ecosystem is challenging in terms of design work, but that once these design skills are acquired by an individual, producing a Py++ library that integrates a C++ library is straightforward and does not require great effort.

Final note

I think it is one of the most important details of Py++ that it does not require great effort to integrate a C++ library. This is the reason why I worked a lot on the transpiler config feature.