Broken: cider-eval-defun-at-point needs a Clojure REPL

[arichiardi]

Expected behavior

It should evaluate forms from cljs files to cljs repl with cider-eval-defun-at-point

Actual behavior

The form is not evaluated with message:

user-error: `cider-eval-defun-at-point' needs a Clojure REPL.
If you don't know what that means, you probably need to jack-in (`C-c M-j').

Steps to reproduce the problem

In replumb, I can provide a branch, lein repl then connect or jack-in then start a Cljs repl with either:

(require 'figwheel-sidecar.repl-api)
(figwheel-sidecar.repl-api/start-figwheel!)
(figwheel-sidecar.repl-api/cljs-repl)

Same thing with cider-jack-in-clojurescript, of course targeting the cljs repl.

Environment & Version information

CIDER version information

;; CIDER 0.11.0 (Bulgaria), nREPL 0.2.12
;; Clojure 1.7.0, Java 1.8.0_72

Emacs version

24.5

Operating system

Ubuntu Linux

Maybe related: clojure-emacs/cider-nrepl#301

[Malabarba]

What is the value of cider-repl-type in your cljs repl buffer?

[arichiardi]

Steps, hopefully right: open ielm, C-c C-b, choose the repl, then cider-repl-type, result clj.

[arichiardi]

@Malabarba By eye bisecting, and then trying, I found that the culprit is 9672dfa

[Malabarba]

choose the repl

Just to be sure. You did choose the ClojureScript repl, right?

then cider-repl-type, result clj.

Then this is the bug, and it's probably on the Clojure side. The commit you found helped exposed the bug, but didn't cause it.

You should have a couple of nrepl-messages buffer around. Please show us the contents of this buffer for the ClojureScript connection.

[arichiardi]

I had chosen the wrong one, I now checked and it is a cljs cider-repl-type. Sorry about that, still the error appears:

Here is nrepl-messages (cut, tell me if you need more):

 (<-- 
  id  "11"
  out  "Prompt will show when Figwheel connects to your application\n"
  session  "6eaf3a92-ca04-49f1-9474-40bbf461152a"
)
(<-- 
  id  "11"
  out  "To quit, type: :cljs/quit\n"
  session  "6eaf3a92-ca04-49f1-9474-40bbf461152a"
)
(<-- 
  id  "11"
  ns  "cljs.user"
  session  "6eaf3a92-ca04-49f1-9474-40bbf461152a"
  value  "nil"
)
(<-- 
  id  "11"
  session  "6eaf3a92-ca04-49f1-9474-40bbf461152a"
  status  ("done")
)
(<-- 
  changed-namespaces  (dict 
                        cljs.analyzer  (dict 
... a lot of stuff
  id  "11"
  repl-type  "cljs"
  session  "6eaf3a92-ca04-49f1-9474-40bbf461152a"
  status  ("state")
)

[alexander-yakushev]

Same issue for me. cider-repl-type in CLJS repl buffer is "cljs", in a .cljs file it's "clj".

[bbatsov]

How many times can we break this? :-)

[Malabarba]

The evidence that's been presented so far seems to indicate there's a problem in cider-map-connections (or one of the functions it calls). Unfortunately, I'm unable to reproduce this issue. =/

Does anyone have simple steps for this?
Even better, if one of you would like to edebug and step through cider-map-connections, that would be a huge help. :-)

[alexander-yakushev]

clojurescript-mode is not automatically enabled for .cljs buffers. When manually enabled, everything works fine again. So, is it a bug with clojure-mode, or my setup is wrong?

[Malabarba]

Oh. Could be something in your setup. Try with emacs -q + package-initialize.

If it turns out to be from your setup, this kind of thing is a little annoying to debug. The best I can think of would be to look through the auto-mode-alist variable, but that's a pretty big variable.

[bbatsov]

Seems fine to me - https://github.com/clojure-emacs/clojure-mode/blob/master/clojure-mode.el#L1495

[alexander-yakushev]

Yes, it's a bug in my setup. Pretty sure it is also the case for @arichiardi .

[bbatsov]

What was the bug in your setup?

[arichiardi]

It can be, I am using emacs live and they might be behind in initializing variables. What did you do @alexander-yakushev?

[alexander-yakushev]

I had custom mode-alist in my (use-package clojure-mode ...) that marked clj[sxc]? as clojure-mode.

[arichiardi]

That's probably the same in emacs live, i guess clojurescript-mode is a new thing?

[arichiardi]

Thanks @Malabarba, while I am going to fix this now, I will check that your warning works 👯

[arichiardi]

I confirm the warning works like a charm, I see it in the Messages buffer. I am now going to fix this.

[bbatsov]

OK, so I guess we can close this. Teamwork for the win! :-)

[arichiardi]

Yes I confirm this works and I patched my emacs-live in order to remove this.

[kthu]

I know this issue is closed, but I'll just put the following up here for future reference;

For me this problem consistently shows up in the latest version of Cider (20160612.1230) if clj-refactor is installed. It worked fine in the previous version (I guess before the workaround in e2f9ffd), but I have struggled with this problem in earlier versions.

I am using Spacemacs where clj-refactor is part of the Clojure layer, but I was able to reproduce the problem starting with an empty .emacs.d after installing Cider and clj-refactor only. Deleting clj-refactor from the elpa/ directory fixes the problem.

This is using Emacs 24.5.1 on Ubuntu 16.04

[Malabarba]

I've deployed a possible fix for this. Let me know if it works for you (make sure lein fetches today's snapshot).

[kthu]

Yes. That fixed it! Thank you.