(I feel obliged to give an example of what I think is a bad developer experience, and I would point at Haskell. There are far too many toolchain variations to choose from, all of which seem to be in active use -- cabal, cabal-v2, stack, cabal-v2+nix, stack+nix; ghcid, ghcide, etc. etc. Once your choices are made and your stack is configured, you'll be okay, except for the inordinately long build times. But in my experience, it can be a real nuisance to come back to a half-finished Haskell project -- say, on a new machine -- unless you've made a singular, personal commitment to a specific technology stack. Leave yourself copious notes, Makefiles, and shell.nixes to remember how to get the thing running! But this is a personal sob-story and a digression.)
My main argument -- which I think is being lost here -- is that Ocaml isn't IDE unfriendly, and especially not so unfriendly that an entirely new language is needed just so that FP and IDEs can happily coexist.
(Well, except for Windows. Ocaml + Windows is more unpleasant than it needs to be.)