Clojurians Log v2
cider
Has anyone seen an error like this before in the nrepl middleware? Not entirely sure what causes it - but I think it happens when there’s an exception and cider tries to print it
Trigger by evaling (throw (ex-info "foo" {}))
hmm looks like it might be a subtle env issue with jdk’s and JAVA_HOME etc… 👀
yeah ok that seems to have fixed it… JAVA_HOME was pointing to the system java rather than the one I was using… slight jabba misconfiguration
hmm odd I’ve just seen the exception again… looks like this hasn’t fixed it after all :thinking_face:
hey guys, has any of you experienced a weird behaviour while using clj-refactor where cider seems to be resolving to the wrong project type ?
I have a deps.edn-based project and it is supposed to handle refactoring commands pertaining to that particular type of project. Though when launching an add-project-dependency for e.g it tells me that it tried to lookup for a :dependencies key which means that somehow it considers the project as a leiningen-based one.
Debugger entered--Lisp error: (search-failed ":dependencies")
re-search-forward(":dependencies")
cljr--insert-into-leiningen-dependencies("mount" "0.1.16")
cljr--add-project-dependency("mount" "0.1.16")
cljr-add-project-dependency(nil)
funcall-interactively(cljr-add-project-dependency nil)
#<subr call-interactively>(cljr-add-project-dependency nil nil)
apply(#<subr call-interactively> cljr-add-project-dependency (nil nil))
(let ((ido-cr+-current-command command)) (apply orig-fun command args))
call-interactively@ido-cr+-record-current-command(#<subr call-interactively> cljr-add-project-dependency nil nil)
apply(call-interactively@ido-cr+-record-current-command #<subr call-interactively> (cljr-add-project-dependency nil nil))
call-interactively(cljr-add-project-dependency nil nil)
command-execute(cljr-add-project-dependency)
@rickmoynihan If this started happening recently, it might be related to the fixes we've tried for https://github.com/clojure-emacs/orchard/issues/103
I'm assuming that something might be missing in your JDK 8 installation. Probably you won't get the error with JDK 11.
@bozhidar yeah I think it’s happened recently since I updated
Mention your issue in the ticket, so Jeff's aware of this.
The Java integration is one of the most frustrating things in CIDER, as things there constantly break...
Now I miss the days where each JDK took 5 years to ship. 😄
Ok, interesting… I could possibly use jdk11, though it’ll mean I need to use two jdks for different services
nevermind, I debugged it using edebug and figured out that it was because I had a pom.xml inside the project (which was generated by clj-new)
(defun cljr--add-project-dependency (artifact version)
(let* ((project-file (cljr--project-file))
(deps (cljr--project-with-deps-p project-file)))
(cljr--update-file project-file
(goto-char (point-min))
(if deps ;; SO WHEN deps IS nil
(cljr--insert-into-clj-dependencies artifact version)
(cljr--insert-into-leiningen-dependencies artifact version)) ;; clj-refactor CONSIDERS THE PROJECT AS A lein-based ONE
(cljr--post-command-message "Added %s version %s as a project dependency" artifact version)
(when cljr-hotload-dependencies
(if deps
(back-to-indentation)
(paredit-backward-down))
(cljr-hotload-dependency)))))
Hello. I am missing a CIDER command: cider-toggle-ignore-next-form. It should either place or remove #_ at current cursor position.
What is the honorable thing to do? Just put it in my custom config? Pull-Request to the Spacemacs configuration layer? Pull-Request to CIDER?
@witek It should like something that should be in clojure-mode, as it doesn't require a REPL to work.
I'm just wondering that's the value of adding a command that adds or delete two characters at point. Probably that'd be more useful if it was applied to the preceding or containing form.
(as it saves you the extra movement)
Any tips for troubleshooting a slow cider-connect? The ;; Connected to nREPL server - <nrepl://localhost:49906>... comment appears immediately, but it takes about 45 seconds for the user> to appear. I’m using cider-nrepl 0.25.9 and the CIDER snapshot <tel:202102131151|20210213.1151> on Emacs 27.1. I start the server with lein repl.
first guess, its the state tracker populating the emacs side with all the known namespaces and vars
@dpsutton that makes sense--this is a large project. Is there any way to pre-calculate this (something we could do in our build)?
my suspicion is its not the calculation of it but just sending to emacs, parsing, and storing in memory. so no good solution off hand that i can think of (if this is even the problem anyways)
i think you can turn this off and you'll just lose a bit of syntax highlighting
wow. i'm not actually seeing a way to not subscribe to this information
I tried enabling debug on quit and hitting C-g during the wait, as the troubleshooting guide that @bozhidar linked to suggests, but I don’t get any backtrace or any other information. so I’m not sure how to confirm your hypothesis. for a given REPL process, if I disconnect and then reconnect, the second connection is instantaneous.
yeah that's really about a hanging process
Yeah, you can't. It should probably be configurable, but it's not as even basic things like the connection type are handled by it.
start profiling, cider connect, wait until it unfreezes and run profiler-report. note do not run profiler-stop. you'd think you start, stop, see results. but you can only see results of actively running profilers.
yeah. i imagine the connection type always stays in there. just not the huge map of dictionaries of nses
i remember someone having an issue with millions of keywords or something like that
surprised one result of that wasn't to turn off the tracking
profiling output in thread
- command-execute 9,655,949 73%
- call-interactively 9,655,949 73%
- funcall-interactively 9,655,949 73%
- counsel-M-x 9,655,949 73%
- ivy-read 9,355,822 70%
- ivy-call 7,208,547 54%
- counsel-M-x-action 7,208,547 54%
- command-execute 7,208,547 54%
- call-interactively 7,208,547 54%
- funcall-interactively 7,208,515 54%
- cider-connect 3,652,581 27%
- cider--update-host-port 2,435,135 18%
- cider-select-endpoint 2,435,135 18%
- cider--ssh-hosts 1,562,797 11%
+ eval 1,542,325 11%
+ seq-map 17,400 0%
+ version< 3,072 0%
+ cider--completing-read-port 373,606 2%
+ cider--completing-read-host 237,411 1%
+ cider--infer-ports 235,481 1%
+ seq-uniq 24,816 0%
split-string 1,024 0%
- cider-nrepl-connect 1,208,190 9%
- nrepl-start-client-process 1,208,190 9%
+ run-hooks 651,591 4%
+ #<compiled 0x1feb8b26da11> 477,204 3%
+ nrepl--init-capabilities 45,358 0%
+ nrepl--init-client-sessions 18,584 0%
+ nrepl-connect 7,181 0%
+ cider--update-project-dir 5,120 0%
- cider--check-existing-session 4,136 0%
+ sesman-current-sessions 4,136 0%
+ profiler-report 3,555,279 26%
profiler-start 655 0%
+ read-from-minibuffer 1,657,843 12%
+ ivy--reset-state 356,434 2%
+ ivy--update-prompt 3,104 0%
+ #<compiled 0x1feb870c8e1d> 1,056 0%
+ counsel--M-x-externs 300,127 2%
+ timer-event-handler 1,626,289 12%
+ nrepl-client-filter 833,280 6%
+ redisplay_internal (C function) 786,459 5%
+ ucs-normalize-hfs-nfd-pre-write-conversion 115,835 0%
+ linum-update-current 75,598 0%
+ ... 45,360 0%
+ ucs-normalize-hfs-nfd-post-read-conversion 41,344 0%
+ eldoc-schedule-timer 584 0%don’t really see any smoking gun here
did you run this report after the 45 second freeze?
maybe run it during the 45 second freeze?
I ran the report after (but it was profiling the whole time). I will try running the report during.
Is there a way to run some snippet just after jack in? Similar to cider-custom-cljs-repl-init-form but just for clj.