Idea: add statistical labeling to elm-test

[Janiczek]

This thread triggered my “coverage + AFL + fuzzing” buzzword bingo sense, so I’ll leave this here in case some folks don’t know about elm-coverage by @ilias https://github.com/zwilias/elm-coverage

[gampleman]

Speaking of AFL, I wonder how feasible it would be to start thinking about this in terms of symbolic execution, i.e. move away (at least partially) from the random aspect of property testing.

I think the advantage here is that Elm is quite a minimalist language, so perhaps this could be a fruitful area of exploration. In particular purity seems to be a huge boon for symbolic execution as does the very limited FFI (i.e. kernel code would have to probably be reimplemented in some logic language to correctly suggest branches, but there is not that much of it).