triple printlns

[borkdude]

I'm seeing thrice the output of println when I enable [cider/cider-nrepl "0.10.0-SNAPSHOT"] in my user profile. I didn't have this problem with 0.9.0.

In case it's useful, I'm using these other plugins in my profiles.clj:

        [jonase/eastwood "0.2.1"]
        [lein-kibit "0.1.2"]
        [lein-ancient "0.6.7"]
        [lein-cloverage "1.0.6"]
        [lein-bikeshed "0.2.0"]
        [lein-try "0.4.3"]
        [lein-plz "0.3.3"]

A look into my project.clj:

  :dependencies [[org.clojure/clojure "1.7.0"]
                 [org.clojure/clojurescript "1.7.122"]
                 [org.clojure/core.async "0.1.346.0-17112a-alpha"]
                 [compojure "1.4.0"]
                 [ring-server "0.4.0"]
                 [mysql/mysql-connector-java "5.1.35"]
                 [org.clojure/java.jdbc "0.3.7"]
                 [lib-noir "0.9.9"]
                 [yesql "0.4.2"]
                 [org.webjars/bootstrap "3.3.4"]
                 [org.webjars/jquery "1.11.1"] ;; needed by bootstrap
                 [org.webjars/jquery-throttle-debounce-plugin "1.1"]
                 [prismatic/schema "0.4.3"]
                 [digest "1.4.4"]
                 [secretary "1.2.3"]
                 [liberator "0.13"]
                 [cljs-http "0.1.35"]
                 [reagent "0.5.1"]
                 [wrap-verbs "0.1.1"]
                 [com.taoensso/timbre "4.0.1"]
                 [environ "1.0.0"]
                 [prismatic/dommy "1.1.0"]
                 [com.draines/postal "1.11.3"]
                 [com.novemberain/pantomime "2.6.0"]
                 [amazonica "0.3.24"]
                 [org.jsoup/jsoup "1.8.2"]
                 [optimus "0.17.1"]
                 [hiccup "1.0.5"]]

  :plugins [[lein-cljsbuild "1.0.6"]
            [lein-immutant "1.2.3"]
            [lein-figwheel "0.4.0"]
            [lein-environ "1.0.0"]
            [lein-auto "0.1.2"]]

[Malabarba]

As per slack discussion, the first suspect is timbre. I'll have to try it out to figure out whether we're doing something wrong, but I'm very busy this week.

[Kungi]

I had a look into this problem and it seems to me that timbre is not the culprit. Even when I replace calls to timbre with println I also get the triple prints.

[Kungi]

It does not triple print when I run my application through the emacs cider repl. Only when I use lein repl.

[Malabarba]

I had a look into this problem and it seems to me that timbre is not the culprit. Even when I replace calls to timbre with println I also get the triple prints.

Did you also remove the timbre dep completely?

It does not triple print when I run my application through the emacs cider repl. Only when I use lein repl.

What is the complete workflow in that case? Is it something like the following?

1. lein repl on the command-line;
2. Evaluate a println on this REPL.

Or is it like this?

1. lein repl on the command-line;
2. cider-connect in Emacs;
3. Evaluate a println in a Clojure buffer.

[Kungi]

I was able to reproduce the bug in a quite minimal example:

https://github.com/Kungi/reproduce-triple-printlns

I run lein repl and type (go) without connecting cider-nrepl to it and open "http://localhost:3333/foo" in the browser and get a triple println.

[Malabarba]

Thank you, this will be hugely useful. I expect I'll be able to fix it over the weekend.

[Malabarba]

Ok, I see what's going on.
The middleware isn't actually bugging out, it's just that, when you start lein repl on a terminal, both the server's *out* and the client's *out* go to the same place (the terminal).
So you see all outputs in one place.

The reason you see 3 outputs instead of 2 is that lein repl apparently creates 2 sessions on the nREPL server, so you see 1 output from the server and 1 from each session. But I have no idea why two sessions are being created.

[Malabarba]

In any case, I think the solution is to disable this automatic forwarding of the server out to every session, and instead only forward it to sessions created by cider.el (via some subscription request). Which is what @bbatsov originally suggested on the PR and I was too lazy to implement. =P

[bbatsov]

I'm in favour of this idea. :-)