Cnx VS bfcpp

The biggest obstacles to make some form of functional programming (think more like Rust than Haskell) possible are __VA_OPT__, auto, lambdas, and defer. IIRC the first two are coming in C23. The last two might make it into the next standard after that, but unlikely to make 23. If you're willing to limit yourself to GNU C (so compiling with GCC or Clang, or maybe ICC) you can get pretty close already today. (shameless self plug) My project Cnx (https://github.com/braxtons12/Cnx) attempts to bring a more modern approach to the language, and a big chunk of the features it implements are functional in nature.

The preprocessor is actually quite fast, as long as you are just doing primitive replacement things. I benchmarked my preprocessor brainfuck interpreter (without optimizations) against a constexpr brainfuck interpreter (without optimizations), and it beat constexpr for interpreting smaller programs. isort4 for example is a brainfuck program that does insertion sort on 45 inputs, and the preprocessor implementation was more than twice as fast as the constexpr one. Larger programs are slower to interpret with the preprocessor, because it always needs to copy the entire program code.

PS: I wrote a bit of an explanation for my preprocessor brainfuck interpreter, maybe you can learn a few tricks from that: https://github.com/camel-cdr/bfcpp/blob/main/TUTORIAL.md

Since you mentioned macros: I wrote a brainfuck interpreter using only the preprocessor, so it interprets brainfuck at compile time. The entire thing is a huge abuse of macros: https://github.com/camel-cdr/bfcpp