Improvements to OCaml code editing: the basics of a refactor engine
tarides.com79 points by nukifw 6 hours ago
79 points by nukifw 6 hours ago
Extract expression is such a common refactoring, happy to see support for it. I'm even more curious to see now if there are going to be more advanced refactoring possible. A List.map, to for loop, and back refactoring would be such a great thing to have in how I program.
Only thing that would go on my OCD nerve, is the lack of an empty newline when the show_markup function is extracted. Kind of "common sense" when writing top level bindings to leave some breathing room between them.
For the first point, you are right. We start by a common (but very useful) feature. Since OCaml allows an infinite level of nesting (and different kind of structure item) it was still a bit of a challenge, mostly for finding the right UX.
For the second point we delay the aeration convention to the formatter (ocamlformat). It can be configure in a different way :)
Thanks for your feedback!
Very cool!
Does it replace identical expressions in the same scope? Like:
let tau = 3.14 +. 3.14
let pi = 3.14
let tau = pi +. pi
EDIT: Or even crazier with function:
let _ = (x + 1) + (y + 1)
let plus_one a = a + 1
let _ = (plus_one x) + (plus_one y)
Nop, for the moment, we try to not "infer user usages"! But if you extract an expression with variable, there will be, obviously, not be repeated:
let tau =
let pi = 3.14 in
pi + pi
let fun_name1 pi =
pi + pi
let tau =
let pi = 3.14 in
fun_name1 piI’ve been working on similar refactoring and grepping utility functions for vim, though mine are meant to work in any language.
The most recent one I’ve made runs 'git grep' on the word under cursor or on the visual selection and puts everything into a quick fix list.
Since it is generic, it works on any phrases as well, and helps me find prose snippets and phrases in docs or other writings.
Tarides has an OCaml focus. Elsewhere on their site, with an interesting omission:
"Examples of functional programming languages include OCaml, Erlang, Clojure, Haskell, Scala, and Common Lisp."
F# and OCaml are close cousins, and for me F# is even cleaner syntax and twice as fast (M4 Mac Studio).
It was a struggle for me to overcome my partisan preference for OCaml.
F# started as an Ocaml port to the CLR. In the end, they decided to port a subset of the language so you get mostly the same syntax, no parametrised modules which is a big part of Ocaml but in exchange you can call libraries from other languages using the CLR. Of course, that comes with the issue of being a second class citizen in the .Net family, albeit a very nice second class citizen.
F# have drifted a bit further since but not that much. It’s a nice language but it’s not really comparable to Ocaml at this point. At least, I don’t see many cases where people would have to chose one or the other. I think people know if they want the CLR or not.
I make little use of .NET; my hand-coded atomic queue is faster than .NET work stealing for parallelism, for example. I like the slightly cleaner syntax and twice as fast for my applications.
I'm escaping the black hole of Haskell. If one can code in Chez Scheme, one can code without parametrized modules, as much as I admire them. And there are also F# additions OCaml could admire.
I held back for all of your reasons, and finally caved. I never saw myself in the .NET ecosystem, but prejudices are a funny thing.
Anyone tried large codebase refactorings using AI?
I feel that Ocaml could really use better VSCode integration more than literally anything else, but the community seems completely oriented around emacs
We try to invest equal time between the OCaml platform (on VSCode) and OCaml-eglot (for Emacs). In fact, I find the VSCode extension to be rich (and generally iso-features with Emacs mode).
However, I (as maintainer of OCaml-eglot) find Emacs easier to extend (the VSCode extension is surprisingly complex when you stray from the beaten path), which seemed perfect for incubating an experimental feature. Furthermore, as mentioned at the end of the article, the feature can also be invoked from a code action, which makes it de facto callable from VSCode once the various PRs have been merged :)
The LSP features, in my experience, work very well in VScode, no shade there. But debugging features like breakpoints still seem to be experimental? Whenever I ask others about it online they usually try to tell me that print debugging is superior anyways
My experience is that everything goes into Merlin, which is Ocaml tool for analysis and refactoring, and is then added to editor from there.
Here significant work was done to properly bridge the gap between Merlin and the LSP server so it should work fine in VSCode.
Generally speaking, the Ocaml community seems to spend a lot of time getting VSCode to work.
[dead]
PRs are mentioned at the end of articles. It is possible to `pin` repositories locally on their different branches to experiment with them locally!
[flagged]