explore property-based testing

[maelle]

https://www.etiennebacher.com/posts/2024-10-01-using-property-testing-in-r/

[maelle]

related to igraph/igraph#2459

[szhorvat]

Google FuzzTest is a framework for doing this for C++ projects. Tamás and I have been looking at it for a while, but realistically, we won't have the time to implement it anytime soon.

The interesting thing about Google FuzzTest is that it doesn't just produce random input. It mutates the input in a random, but also targeted way to try to maximize code coverage (like in coverage-guided fuzzing). Are there similar tools for R that can go beyond dumb random input generation?

BTW many of the tests we have are already like this in spirit, except that they run for a fixed seed. If we allow a variable seed, then we need: (1) an easy way to run the test for the same seed (2) a way to report the seed upon failure. That would have needed a bit more work for the custom testing framework we had in C.