[Clojure] [ANN] nrepl.el 0.1.4 released

I get the the following error when trying to auto-complete e.g. "(pri":

--8<---------------cut here---------------start------------->8---
java.lang.NullPointerException: null
at clojure.lang.Reflector.invokeNoArgInstanceMember (Reflector.java:314)
clojure.stacktrace$print_stack_trace.invoke (stacktrace.clj:51)
clojure.stacktrace$print_stack_trace.invoke (stacktrace.clj:49)
user$eval6701.invoke (nrepl-server.clj:2)
clojure.lang.Compiler.eval (Compiler.java:6465)
clojure.lang.Compiler.eval (Compiler.java:6431)
clojure.core$eval.invoke (core.clj:2795)
clojure.main$repl$read_eval_print__5967.invoke (main.clj:244)
clojure.main$repl$fn__5972.invoke (main.clj:265)
clojure.main$repl.doInvoke (main.clj:265)
clojure.lang.RestFn.invoke (RestFn.java:1096)
clojure.tools.nrepl.middleware.interruptible_eval$evaluate$fn__768.invoke (interruptible_eval.clj:57)
clojure.lang.AFn.applyToHelper (AFn.java:159)
clojure.lang.AFn.applyTo (AFn.java:151)
clojure.core$apply.invoke (core.clj:600)
clojure.core$with_bindings_STAR_.doInvoke (core.clj:1769)
clojure.lang.RestFn.invoke (RestFn.java:425)
clojure.tools.nrepl.middleware.interruptible_eval$evaluate.invoke (interruptible_eval.clj:42)
clojure.tools.nrepl.middleware.interruptible_eval$interruptible_eval$fn__809$fn__811.invoke (interruptible_eval.clj:170)
clojure.core$comp$fn__3758.invoke (core.clj:2276)
clojure.tools.nrepl.middleware.interruptible_eval$run_next$fn__802.invoke (interruptible_eval.clj:137)
clojure.lang.AFn.run (AFn.java:24)
java.util.concurrent.ThreadPoolExecutor$Worker.runTask (ThreadPoolExecutor.java:886)
java.util.concurrent.ThreadPoolExecutor$Worker.run (ThreadPoolExecutor.java:908)
java.lang.Thread.run (Thread.java:662)
--8<---------------cut here---------------end--------------->8---

I use tools.nrepl 0.2.0-beta9 and clojure-complete 0.2.2

The nREPL server is started with this script (nrepl-server.clj as seen
in stack-trace):

--8<---------------cut here---------------start------------->8---
(use '[clojure.tools.nrepl.server :only [start-server stop-server]])

(defonce server (start-server :port 4005))
(println "nREPL server started at port 4005")
--8<---------------cut here---------------end--------------->8---


-Stefan

Tim King <kin...@...com> writes:

Hey,

Still a fairly newbie emacs user. But when trying to evaluate a form with *C-x
C-e ( *after* M-x nrepl-jack-in )*, I still get this error (was happening
with nrepl 0.1.3):

*Symbol's function definition is void: lisp-eval-last-sexp*


This call looks like it's used in *
~/elpa/clojure-mode-1.11.5/clojure-mode.el*. I'm using GNU Emacs 24.0.92.1.
I also still have these directories lying around. Should something be blown
away, or uninstalled? Does clojure-mode need to be updated ( 1.11.5 looks
like the latest )?

~/.emacs.d/swank/
~/.emacs.d/elpa/nrepl-0.1.3/
~/.emacs.d/elpa/nrepl-0.1.4/



Thanks

Tim Washington
Interruptsoftware.ca
416.843.9060

Timothy Washington writes:

Hi Timothy,
Hm, I think that's a SLIME function, and clojure-mode currently uses a
bundled SLIME that is compiled when you do clojure-jack-in. With
nrepl.el, you don't do that anymore.

But anyway, once you've nrepl-jack-in'ed all (or the clojure buffers of
that project?) should have nrepl-interaction-mode enabled, and that
binds C-x C-e to nrepl-eval-last-expression.

Does the buffer with the clj file have nrepl-interaction-mode enabled?
You can see that by the nREPL lighter in the mode line or by doing C-h v
nrepl-interaction-mode RET in that buffer. It should say the value is
t.
But not this one.

Bye,
Tassilo

Hey Tassilo, thanks for responding.

Well after *M-x nrepl-jack-in*, I notice that the "*Starting nREPL server...
*" message never goes away. But I don't *think* that's the main problem.

*1)* That help window looks like the following. So it doesn't look like the
*nrepl-interaction-mode* is enabled on emacs startup.


*-UU-:----F1 neuralnet.clj Top (16,0) Git:master (Clojure Fill)
----...*
nrepl-interaction-mode is a variable defined in `nrepl.el'.
Its value is nil

Automatically becomes buffer-local when set in any fashion.

Documentation:
Non-nil if nREPL-Interaction mode is enabled.
Use the command `nrepl-interaction-mode' to change this variable.



*2)* After executing the emacs command '*nrepl-interaction-mode*', I could
then execute *C-x C-e*, but got the following message

No buffer named *nrepl-connection*



*3)* So I went ahead and fired up *M-x nrepl*, and supplied the host and
port. Again, the result message "*Connecting to nREPL on localhost:57260...*"
never goes away (hopefully not the real problem). After this, trying to
execute *C-x C-e* or *C-x C-l* , gives me the error:

Wrong type argument: stringp, nil




Tim Washington
Interruptsoftware.ca
416.843.9060

I think, it is. Normally, after that message it takes ~10secs and then
a *nrepl* buffer should pop up. If that doesn't happen for some reason,
then there's something wrong.
It's enabled automatically in all clojure buffers as soon as nrepl
successfully started, which doesn't seem to happen for you.
Hm, hm, why doesn't it want to connect...

What you could do to debug the issue is open nrepl.el, goto the function
definition of nrepl-jack-in, place point (the emacs cursor) on the
function name, do `M-x edebug-defun RET', and then `M-x nrepl-jack-in
RET'. Now you'll see an arrow in the margin indicating the flow of
control. With `n' you can step one expression further, and the results
are printed in the echo area. What are those values? Especially the
result of `start-process-shell-command' might give a clue.

Bye,
Tassilo

Oooh, this is interesting... Well the code block is below, and you can see
where the arrow is stopping on execution.

;;;###autoload





(defun nrepl-jack-in (prompt-project)
(interactive "P")
*=>*(let* ((cmd (if prompt-project





(format "cd %s && %s" (ido-read-directory-name "Project:
")
*nrepl-server-command*)
nrepl-server-command))
(process (start-process-shell-command
"nrepl-server" "*nrepl-server*" cmd)))
(set-process-filter process 'nrepl-server-filter)
(set-process-sentinel process 'nrepl-server-sentinel)
(set-process-coding-system process 'utf-8-unix 'utf-8-unix)
(message "Starting nREPL server...")))



That form has several calls, the first result is nil. The second result
error's out with the message below. And this is where execution just dies
and goes away

- nil
- *Symbol's value as variable is void: nrepl-server-command*


I can see in nrepl.el where nrepl-server-command is defined. But I don't
know enough about elisp to say what the problem is.


Tim Washington
Interruptsoftware.ca
416.843.9060

Uh, now that's really strange... It means that nrepl.el hasn't been
loaded yet. `nrepl-jack-in' is an autoloaded function meaning once you
call it, nrepl will be loaded too. That contradicts each other...

I suspect you have some old version laying around somewhere. Please do
`M-x list-load-path-shadows RET' and check if there's some entry like

/foo/bar/nrepl hides ~/.emacs.d/elpa/nrepl-0.1.4/nrepl

If that's not the case, maybe there's something broken in your configs
or some addon. To rule that out, please start emacs using "emacs -Q".
Then copy this code into the *scratch* buffer and evaluate it.

--8<---------------cut here---------------start------------->8---
(progn
(add-to-list 'load-path "~/.emacs.d/elpa/clojure-mode-1.11.5")
(add-to-list 'load-path "~/.emacs.d/elpa/nrepl-0.1.4")
(require 'nrepl)) ;; <== C-x C-e here!
--8<---------------cut here---------------end--------------->8---

Then open some clojure file in some leiningen project and do `M-x
nrepl-jack-in RET'. Does it work then?

Bye,
Tassilo

Heyya, thanks for your help so far on this. See comments inlined...


*
*

*No Emacs Lisp load-path shadowings were found*



hanging "Starting nRepl server..." message, when executing M-x
nrepl-jack-in. *



*The **"Starting nRepl server..." message is still hanging. **But I ran the
emacs debugger on nrepl, and this time, it got past the "
nrepl-server-command" command. The debugger seems to have completed the
entire "nrepl-jack-in" function. I'm genuinely at a loss. *


Bye,
*Thanks again, and see you. *
*
*
*Tim *

Ok, so far so good.
Ok.
Hm, if that function is completed, it should eventually call the process
sentinel function `nrepl-server-sentinel'. Try do edebug that one. Is
it ever reached?

If not, what's the contents of the " *nrepl-server*" buffer?

Bye,
Tassilo

Ok, I got it. It was very interesting. The debugger never got to "
nrepl-server-sentinel". But the *nrepl-server* buffer said:

REPL started; server listening on localhost port 54922
Exception Unsupported option(s) supplied: :headless clojure.core/load-libs
(core.clj:5266)
clojure.core=>



The *nrepl-server-command* function in nrepl.el, has that :headless option
to a repl option, of what I assumed was a *lein* command.

(defcustom nrepl-server-command
(if (or (locate-file nrepl-lein-command exec-path)
(locate-file (format "%s.bat" nrepl-lein-command) exec-path))
*(format "%s repl :headless" nrepl-lein-command)




*
* (format "echo \"%s repl :headless\" | $SHELL -l" nrepl-lein-command))*
"The command used to start the nREPL via nrepl-jack-in.





For a remote nREPL server lein must be in your PATH. The remote





proc is launched via sh rather than bash, so it might be necessary





to specific the full path to it. Localhost is assumed."
:type 'string
:group 'nrepl-mode)



I dug around a bit and found this
issue<https://github.com/technomancy/leiningen/issues/715>.
That was the clue I needed to solved the problem. I have several lein
executables in my path. I made lein.2 the default lein, and now I can "*M-x
nrepl-jack-in*", and execute "*C-x C-e*". So everything works. Thanks very
much for the help. I think this is another step on the road to me being
emacs-ified.


Tim Washington
Interruptsoftware.ca
416.843.9060