Thanks for the link, will read soon.

Agree with everything you said, and I'd also add polymorphic variants. I too am a Haskeller now writing mostly Ocaml. I view the two languages as cousins. Both have great features the other doesn't. Both are worth learning and using.

Interestingly, that's also a disadvantage since you cannot control effects (in the user-supplied code) if you're e.g. writing a library API with callbacks. On the advantage side you have what you mentioned, namely "benign" effects -- but I've found that the "sprinkle a few debug printfs" use case is adequately handled with Debug.Trace... unless you're trying to debug a terminal application! :)

Logging is another popular example of a "benign" effect, but I personally don't think is: I mean, your logging library could be doing all kinds of weird things like shipping logs over TCP to a remote syslog daemon... and I don't want that to be hidden under the rug (using unsafePerformIO or similar) only to blow up later when the network happens to be down.

Anyway, having programmed in O'Caml for 3-4 years, I was won over to Haskell by the compiler-supported (side) effect tracking. I think it's also important to note the distinction between side effects and plain (desired) effects. Not enough people do this in these sorts of discussions.

I know Standard ML, and yes, it has eager evaluation and modifiable references. I've learned to appreciate lazy evaluation and purity, so those have little value to me. I miss parameterized modules, as Haskell's modules are strictly for namespaces. But I'm pretty clueless on the delta from SML to modern OCaml, so maybe I've missed out on valuable developments.