:repl-options :init seems to be ignored by lein repl :headless

[mjwillson]

$ cat ~/.lein/profiles.clj 
{:user
  {
    :plugins [
      [lein-midje "2.0.0-SNAPSHOT"]
     ]
    :repl-options {
      :init (do
         (print "hello")
         (set! *print-length* 20)
         (use 'clojure.stacktrace)
      )
      :timeout 90000
    }
  }
}

If I do a lein repl the init code runs (the 'hello' visible, print-length change has taken effect), whereas if I do a lein repl :headless and connect to it via lein repl :connect, the init code doesn't run, in particular my print-length setting doesn't take effect, which is the main goal here, I want to stop emacs / nrepl.el hanging nastily if I accidentally type some big lazy sequence into the repl.

(Also, neither lein repl nor lein repl :headless seems to run the init code if I run it outside of a folder with a project.clj. Was this intentional? It would be nice if my profiles.clj applied everywhere)

[mjwillson]

Should have said: this is using Leiningen 2.0.0-preview10

[mjwillson]

Anyone have any ideas on this? it's a bit of a drag having to manually enable basic conveniences like stacktrace, pprint, print-level etc, in every new REPL (affecting emacs nrepl.el and other IDEs which use a headless repl)

[technomancy]

Hoping to get this fixed in the next release. In the mean time you should be able to just add those kinds of conveniences to the :user profile's :injections key and they will be applied to every project invocation, not just repls.

[technomancy]

The problem here is that the :init stuff is currently handled by the reply client, so when we don't spin up a client we lose it. We need to support that stuff server-side instead. Need to add this into both the start-server function and the trampoline-repl function. Shouldn't be too tricky for a new contributor if anyone wants to give it a shot.

[kingtim]

I am going to take a crack at this.

[kingtim]

I have a pull request that fixes this for the lein repl :headless case, which I will be submitting shortly.

Trampoline is a slightly different story.
lein trampoline repl seems to be trying to handle this already, but seems to be bombing when I have :repl-options :init in my profiles.clj. Seems like a bug somewhere.. possibly on the repl-y side, have not tracked it down yet.

$ cat ~/.lein/profiles.clj 
{:user
  {
    :repl-options {
      :init (do
         (set! *print-length* 20)
      )
    }
  }
}
$ lein trampoline repl
REPL-y 0.1.0-beta11
Clojure 1.4.0
CompilerException java.lang.RuntimeException: Unable to resolve symbol: ... in this context, compiling:(NO_SOURCE_PATH:1) 
reply.eval-modes.nrepl=> Bye for now!

$ mv ~/.lein/profiles.clj /tmp
$ lein trampoline repl
REPL-y 0.1.0-beta11
Clojure 1.4.0
    Exit: Control+D or (exit) or (quit)
Commands: (user/help)
    Docs: (doc function-name-here)
          (find-doc "part-of-name-here")
  Source: (source function-name-here)
          (user/sourcery function-name-here)
 Javadoc: (javadoc java-object-or-class-here)
Examples from clojuredocs.org: [clojuredocs or cdoc]
          (user/clojuredocs name-here)
          (user/clojuredocs "ns-here" "name-here")
user=> 

[technomancy]

Thanks for the fix! For trampolining it looks like the problem here is that options-for-reply is ignored in trampoline-repl.

[kingtim]

I'll take a look at the trampoline and and address that in a separate PR.

[technomancy]

Hoping to do a release candidate soon; should I plan on this making it in or should we push this off to the next release?

[kingtim]

Apologies for the delay. I am planning to try to get you a fix this weekend. When are you trying to do RC?

[technomancy]

No problem; it's probably a week out at this point. Thanks!

[kingtim]

I spent some time with this over the weekend and here are my findings.

lein trampoline repl is correctly passing the init options to the repl:

$ cat ~/.lein/profiles.clj
{:user { :repl-options { :init (def ^:dynamic *earmuff* 10) } } }
$ lein trampoline repl
Warning: *earmuff* not declared dynamic and thus is not dynamically rebindable, but its name suggests otherwise. Please either indicate ^:dynamic *earmuff* or change the name. (NO_SOURCE_PATH:1)
REPL-y 0.1.0-beta11
Clojure 1.4.0
    Exit: Control+D or (exit) or (quit)
Commands: (user/help)
    Docs: (doc function-name-here)
          (find-doc "part-of-name-here")
  Source: (source function-name-here)
          (user/sourcery function-name-here)
 Javadoc: (javadoc java-object-or-class-here)
Examples from clojuredocs.org: [clojuredocs or cdoc]
          (user/clojuredocs name-here)
          (user/clojuredocs "ns-here" "name-here")
user=> *earmuff*
10

If I set *print-length* in the init options however, somewhere in the pipeline the init form is being abbreviated with ellipses e.g. (ns-java-methods ns))) ...)))
(do [init-forms-elided] (concat special-forms (namespaces ns) (ns-vars ns) (ns-classes ns) (ns-java-methods ns))) ...))) (catch java.lang.Exception f__1852__auto__ (clojure.core/intern (clojure.core/create-ns (quote complete.core)) (quote completions) (clojure.core/fn [prefix__1853__auto__ ns__1854__auto__] [])) (clojure.core/println "Unable to initialize completions."))))) ...) nil (do (set! *print-length* 20)) nil)

which of course causes the the repl startup to fail:

$ lein trampoline repl
REPL-y 0.1.0-beta11
Clojure 1.4.0
CompilerException java.lang.RuntimeException: Unable to resolve symbol: ... in this context, compiling:(NO_SOURCE_PATH:1) 
reply.eval-modes.nrepl=> Bye for now!

This appears to be an issue on the reply side of things and I plan to log it against that.
To me it seems like this particular issue could be resolved, but its up to you how you'd like to handle that.

[technomancy]

OK, my bad; not sure what I was thinking earlier. Thanks for tracking this down. I think you're right that the metadata issue needs to be addressed upstream in reply.

[kingtim]

I submitted trptcolin/reply#89 for the reply issue.