Blog posts about the successor to C/C++ are popular now. However I would prefer to continue to use C/C++ just long enough to implement a better language, in my case ISLisp. So this post isn't about using some new programming language, but about how to harden the core code of Easy-ISLisp. After some discussion with the project maintainer, I think I have reached the exact same conclusion as when I was playing with OpenCOMAL.
I spent some time considering Embedded C++. There were some benefits, especially the first-class string support. However, this introduces a high risk that future changes would go beyond the "Embedded" subset, reducing maintainability by volunteers.
So we're back at C, and standard UNIX libraries like curses (no need to cater for Win32). Cii (e.g. for the C++ features I miss) and Nana proved useful in the past, so they could be integrated too in the medium term. This is pretty-much the exact design I ended up with for OpenCOMAL. So there must be something to it. The difference is, ISLisp is a far more modern programming language than COMAL (or is it more that Lisp is timeless ...).