Port / Rewrite core parts of experiment code in C++

[olafmersmann]

Proposal

Rewrite the core parts of the experiment code in C++11 or maybe even C++17.

Rational

It's been almost 10 years since we rewrote the experiment code in our funky OO dialect of C. Since then, C++ has become much more ubiquitous and there are high quality C++11/17 compilers for all relevant platforms. There are more people who are familiar with "standard" C++ than our particular style of OO-C which means we might attract more contributors.

Benefits

• Easier to understand code structure
• Cleaner code because of better abstractions (std::string vs. char *, ...)
• More opportunities for optimization (C++ compilers are good at devirtualizing / inline across objects/classes, something C compilers really don't do).

Challenges

• Need to maintain backward compatability (i.e. continue exposing C API).
• More complex build setup because we need to detect a (sane) C++ compiler
• Less portable binarys (but the C++ ABI is pretty stable these days).
• Likely longer compile times

[olafmersmann]

During our meeting on 2023-01-18 the general consensus was that we should indeed investigate a C++ port.

Some remarks:

• There is at least one major feature (linear logging - @ttusar) that needs to be merged before we start.
• We need to check platform availability of the different C++ standard (C++11 should be fine, but maybe we can get C++14 or C++17)
• Need to update and simplify language bindings at the same time.