Ulysse

Thanks for your report @correnson, it would be very useful to have a reproduction. It does look related to https://github.com/ocaml/ocaml-lsp/pull/1529 (cc @Tim-ats-d) @correnson if that's indeed the case, you can...

Thank you for taking the time to report that @correnson. That sounds very legitimate. I would even suggest having these code lenses opt-in for the editors that won't benefit from...

Thank you @lessp, that's an interesting proposition. To clarify a bit, this PR does two things: 1. Add `:` as a trigger character for completion 2. Use Construct results to...

> Another experiment I did was to auto-complete valid constructors that return the type in question. Note that this is already an option of the Construct query in Merlin, not...

Thanks for your report @avysk The "output" part of the message doesn't look like the actual error, it "just" shows warnings. Could you share the whole log in `~/.opam/log/ocaml-lsp-server-38378-693842.out` ?

Thanks a lot @davesnx ! Just a quick note: the latest merlin lib for 4.14 is 4.19-414, not 4.18 :-) Also you probably want to revert the CI workflow changes.

I looked at the issues in this PR and finally decided to take the merge route instead in #1577. Thanks for doing that work @davesnx !

Thanks for the very thorough report ! I will have a look.

> Foo.1 is correct in the sense that Bar is Foo.1 according to foo.mli, but Foo.1 in foo.ml is the constructor Foo, which is what Merlin incorrectly returns. This is...

Given how locate works right now, I would expect the correct result of locating `Bar` in the failing test to be `1:9 in foo.mli` since only the `module type of...