Request: variable to control cljr--clj-context-p popup

[eggsyntax]

Hi, let me first say that I find clj-refactor incredibly helpful and make use of it every day. Thanks very much for such a great tool!

I have a fairly minor request: clj-refactor.el contains the following function:

(defun cljr--clj-context-p ()
  "Is point in a clj context?"
  (or (cljr--clj-file-p)
      (when (cljr--cljc-file-p)
        (if (cljr--point-in-reader-conditional-p)
            (cljr--point-in-reader-conditional-branch-p :clj)
          (string-equal (cljr--prompt-user-for "Language context at point? "
                                               (list "clj" "cljs"))
                        "clj")))))

Unlike pretty much everything else in clj-refactor, I find this more annoying than helpful; I nearly always escape out of the popup -- often after typing half the fn/var name I want, meaning I have to retype it once I realize I'm in the popup. I've taken a look to see if I saw a variable that controls this, but I haven't found one.

Can we have a user-settable variable that controls this, such that the popup is skipped if the variable is set to false? It could result in the same behavior that escaping out of the popup does (even if there's a bit of cost to clj-refactor's ability to help out in that case). Of course I'm imagining it set to true by default, so that there's no behavior change unless the user explicitly sets the variable in their user config.

I'm happy to try to put together a PR if that would be helpful.

Thanks!

[expez]

Hi

Sorry you're having a bad time! Unfortunately, the cljs/cljc support isn't that great because I (and I think @benedekfazekas too) only write clj at work. This means we're especially grateful when some of you speak up! :)

Looking at the function one layer out:

(defun cljr--get-aliases-from-middleware ()
  (when-let (aliases (cljr--call-middleware-for-namespace-aliases))
    (if (cljr--clj-context-p)
        (gethash :clj aliases)
      (gethash :cljs aliases))))

I think perhaps we're doing the wrong thing here. In my eyes we should probably:

1. Look for clj context
2. If not found, look for cljs context
3. If not found fall back to that prompt you've come to hate.

Does that sound better to you?

If this doesn't solve the problem then I'm OK with adding a variable too. The current situation sounds awful :(

I'm happy to try to put together a PR if that would be helpful.

That would be lovely!

Thanks for caring!

[eggsyntax]

Oh, no, don't get me wrong! It's FAR from awful, it's just a minor annoyance. Barely even puts a dent in the huge value I get from clj-refactor :)

I've only had time to take a quick look back, but -- offhand, I'm not sure your approach would solve this problem. It only happens in cljc files (which I see I completely failed to mention in my OP). The code handles the case when we're inside a reader conditional, but being outside of reader conditionals triggers the prompt. So I'm not sure it's possible to resolve to clj or cljs context (unless there's a strategy I'm unaware of).

[expez]

I see. So we need a variable that disables the magic require feature when the dialect can't be contextually resolved (to avoid prompting the user). Is that right?

That variable is going to be hard to name!

[eggsyntax]

Yeah, that's what I was thinking.

bypass-context-popup, maybe?
or... bypass-dialect-popup?

I'm definitely hoping to find time to get a PR together, but it'll probably be 1-2 weeks; don't hesitate to change it before then if you get inspired.

Thanks again!