Introduce threading refactoring commands

[rrudakov]

All related tests from clojure-mode pass except of those which "unjoin" lines, in clojure-mode it's implemented via text properties and doesn't work for the same expression after restarting Emacs. If it's an important feature, we can deal with that later.

I also fixed a couple of small issues with indentation and aligning forms.

---

Before submitting a PR mark the checkboxes for the items you've done (if you
think a checkbox does not apply, then leave it unchecked):

• The commits are consistent with our contribution guidelines.
• You've added tests (if possible) to cover your change(s). Bugfix, indentation, and font-lock tests are extremely important!
• You've run M-x checkdoc and fixed any warnings in the code you've written.
• You've updated the changelog (if adding/changing user-visible functionality).
• You've updated the readme (if adding/changing user-visible functionality).

Thanks!

[rrudakov]

Weird, I didn't close it. I guess GitHub did it somehow :)

[bbatsov]

Shouldn't this raise some error if the form at point is not threadable?

[rrudakov]

In clojure-ts-mode it just returns nil, I did the same here. Maybe because it's used in a loop in both thread-first-all and thread-last-all?

[bbatsov]

Much of the refactoring code was copied from clj-refactor in bulk early on, so I didn't pay close attention to each command there - that's why I notice some stuff for the first time.

I think it probably makes sense to have an optional param controlling whether to raise an error or return nil, and perhaps even introduce some lower-level function that's not actually command that all of the commands call. (extracted from clojure-ts-thread)

Any rate - what you're doing right now is a great opportunity for us to examine closely each refactoring command and make them slightly better here and there in the process of porting them to clojure-ts-mode.

[rrudakov]

I agree, I already discovered a few issues in clojure-mode commands and fixed them in clojure-ts-mode implementation.

I updated this function, it signals user-error now if called interactively and there is no threading form.

[rrudakov]

I think it probably makes sense to have an optional param controlling whether to raise an error or return nil, and perhaps even introduce some lower-level function that's not actually command that all of the commands call. (extracted from clojure-ts-thread)

Perhaps we could look into it when we port more refactoring commands and see some pattern that can be extracted.

[bbatsov]

All related tests from clojure-mode pass except of those which "unjoin" lines, in clojure-mode it's implemented via text properties and doesn't work for the same expression after restarting Emacs. If it's an important feature, we can deal with that later.

I wouldn't say that's particularly important.