Figwheel node process easily hangs and must be restarted

[cjohansen]

I'm using figheel.main with node, and the process locks and needs to be restarted quite frequently.

To reproduce, start a figwheel-main REPL, using this config:

^{:watch-dirs ["src"]
  :open-file-command "emacsclient"}
{:main "myapp.dev"
 :optimizations :none
 :pretty-print true
 :target :nodejs
 :source-map true
 :npm-deps {:puppeteer "1.8.0"
            :minimist "1.2.0"}
 :install-deps true}

Then do this:

(def http (js/require "http"))

(http.request
 {:host "somewhere.com"
  :path "/api/hello"
  :method "POST"
  :headers {"Content-Type" "application/json"}}
 (fn [res]
   (println res)))

Now, as you can see I forgot to wrap my parameters in (clj->js), and so I'm passing an opaque object to http.request, which causes it to fail, and then the process freezes. The log says:

 [Figwheel REPL] Session ID: 50e09152-e516-40f8-b835-af2593675710
 [Figwheel REPL] Session Name: Elmo
nhandled 'error' event
      ^

Error: connect ECONNREFUSED 127.0.0.1:80
    at Object._errnoException (util.js:1024:11)
    at _exceptionWithHostPort (util.js:1046:20)
    at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1182:14)

At this point, the REPL refuses to evaluate anything, even simple (js/console.log "Hello") just hangs. If I save files, then figwheel sees that and starts compiling, but never finishes. Save another file, and it'll print "[Figwheel] Compiling build dev to "target/node/dev/dev-main.js" again, but it never completes and the REPL is effectively dead.

Verified on both 0.1.9 and 0.2.0-SNAPSHOT.

[cjohansen]

I was about to mention that this is not the only way to trigger this problem - any node error will do. And then it hit me - Figwheel does not set up uncaughtException or unhandledRejection handlers, so any runtime error will kill the REPL. I suggest that Figwheel adds those handlers to provide a better developent experience.

[arichiardi]

I have noticed this too and it is definitely confusing (especially in cider)