Case in point:
Now, granted, modular implicits are a complex feature that requires substantial expertise to be qualified to contribute to, but considering even dynamic arrays took 6 years, it sometimes feels like any change to the OCaml language requires this level of investment.
And as mentioned in the post, a language being slow in its evolution isn’t a huge problem for me in itself, provided there are reasonable means for developers to work around these limitations, namely, suitable metaprogramming and build customisability, but in both regards here OCaml’s offerings are somewhat lacklustre (metaprogramming being ppxlib, and build customisability being dune and it’s inflexibility).
As a case in point, recently while trying to write some FFI bindings for Lean to the Godot Game engine, over maybe 30/40 minutes of hacking I was able to implement my own incomplete and adhoc in-house ctypes implementation specialised to the conventions of the engine, and the whole time I could only thank myself I hadn’t chosen to try to do this in OCaml, where it would have been substantially more difficult and far more brittle.