CIDER functionality does not work for code in JARs anymore

[hanshuebner]

Expected behavior

C-c t v on a function that lives in a jar file should toggle tracing, like it does for functions defined in local files.

Actual behavior

Error message is shown:

‘cider-toggle-trace-var’ requires the nREPL op "toggle-trace-var".  Please, install (or update) cider-nrepl 0.18.0 and restart CIDER

Steps to reproduce the problem

M-. to the source code of a function that is defined in a .jar file, then type C-c t v on the function name. Alternatively, find a clojure function that is call from the source file loaded from the jar, and press M-. on it, this yields the message

‘cider-find-var’ requires the nREPL op "info".  Please, install (or update) cider-nrepl 0.18.0 and restart CIDER

Environment & Version information

CIDER version information

;; CIDER 0.18.0 (Saigon), nREPL 0.2.13
;; Clojure 1.8.0, Java 1.8.0_51

Lein/Boot version

Leiningen 2.8.1 on Java 1.8.0_51 Java HotSpot(TM) 64-Bit Server VM

Emacs version

GNU Emacs 25.2.1 (x86_64-unknown-linux-gnu) of 2017-06-30

Operating system

CentOS 7

[bbatsov]

Hmm, that's pretty weird. I'd expect for you to get some missing sesman-session error in such a case, but I can't imagine why the middleware'd be missing. That seems relevant to what we're trying to fix in #2483. //cc @vspinu

[vspinu]

Seems to be unrelated to session management. In the current master sessions are auto-linked to jars. That error might be a cover-up irrelevant to the core-issue. What does your *nrepl-messages* buffer say?

For me cider-toggle-trace-var either time-outs when called on a clojure.core var or throws (wrong-type-argument stringp nil), so something is clearly amiss in there.

[hanshuebner]

@vspinu I don't have a *nrepl-messages* buffer, maybe because I'm using M-x cider-jack-in? Let me know how I can assist if I can.

[vspinu]

(setq nrepl-log-messages t)

[hanshuebner]

When I type `M-.` in a source file to navigate to the implementation of `map?`, this is written to the `*nrepl-messages ...` buffer: ``` (<-- id "19" session "b641abbb-ba5d-4dd4-8d70-740ed8737e2b" time-stamp "2018-10-12 20:59:34.526307859" added "1.0" arglists-str "[x]" column 1 doc "Return true if x implements IPersistentMap" file "jar:file:/home/hans/.m2/repository/org/clojure/clojure/ 1.8.0..." line 167 name "map?" ns "clojure.core" resource "clojure/core.clj" see-also ("clojure.core/hash-map" "clojure.core/set?" "clojure.core/vector?") static "true" status ("done") ) ``` When I then use `M-.` inside of the code from the jar, nothing else shows up in `*nrepl-messages ...`.

[kommen]

Just throwing it in here, but this could also be an issue since #2467

[kommen]

Although cider-toggle-trace-var just works for me for defns in files inside jars...

[jeremyheiler]

This issue:

user-error: ‘cider-find-var’ requires the nREPL op "info".  Please, install (or update) cider-nrepl 0.18.0 and restart CIDER

Also appears to exist for source files that are not within the directory that the REPL was started in.

That is, I have a project.clj that has additional source paths from a "cousin" directory.

[bbatsov]

I'm pretty sure that's a regression from #2467.

[bbatsov]

(someone should debug what's happening in the cider-ensure-nrepl-op function)

[jeremyheiler]

I did some preliminary debugging, and tracing through these functions:

cider-nrepl-op-supported-p
cider-current-repl
cider-repls
sesman-current-session

The call to (sesman-current-session 'CIDER) returns nil.

[vspinu]

Can you step through sesman-friendly-session-p and see what's going on? Looks like your jar is not in the classpath. Should be trivial to track down. Path mismatch of some kind.

[kommen]

For what it is worth, I just tried to reproduce the issue, but toggling tracing with cider-toggle-trace-var works for me.

;; CIDER 0.19.0snapshot (package: 20181019.812), nREPL 0.4.5
;; Clojure 1.9.0, Java 1.8.0_102

Navigated to a clj source file of a maven dependency of a project using deps.edn, and in the clj file inside the jar cider-toggle-trace-var worked as expected.

[jeremyheiler]

I believe this commit has fixed the issue for me: e9108a0.

Thank you @vspinu

[vspinu]

Right. This one should be fixed. @hanshuebner, @kommen could you please confirm?