<< Back to previous view

[CLJ-1944] (into {}) fails for pairs represented as anything other than vectors Created: 03/Jun/16  Updated: 05/Jun/16  Resolved: 05/Jun/16

Status: Closed
Project: Clojure
Component/s: None
Affects Version/s: Release 1.7
Fix Version/s: None

Type: Defect Priority: Major
Reporter: John Napier Assignee: Unassigned
Resolution: Declined Votes: 0
Labels: bug, compiler, exceptions

Linux 3.13.0-63-generic #103-Ubuntu SMP x86_64 GNU/Linux


This works:

(into {} [[:a 1]])
;=> {:a 1}

This also works:

(into {} (list (vector :a 1)))
;=> {:a 1}

Bizarrely enough, even this works:

(into {} [{:a 1}])
;=> {:a 1}

This produces a ClassCastException:

(into {} [(list :a 1)])
;=> java.lang.ClassCastException: clojure.lang.Keyword cannot be cast to java.util.Map$Entry
	at clojure.lang.ATransientMap.conj(ATransientMap.java:44)
	at clojure.lang.ATransientMap.conj(ATransientMap.java:17)
	at clojure.core$conj_BANG_.invokeStatic(core.clj:3257)
	at clojure.core$conj_BANG_.invoke(core.clj:3249)
	at clojure.lang.PersistentList.reduce(PersistentList.java:141)
	at clojure.core$reduce.invokeStatic(core.clj:6544)
	at clojure.core$into.invokeStatic(core.clj:6610)
	at clojure.core$into.invoke(core.clj:6604)
	at user$eval4419.invokeStatic(form-init625532025826918014.clj:1)
	at user$eval4419.invoke(form-init625532025826918014.clj:1)
	at clojure.lang.Compiler.eval(Compiler.java:6927)
	at clojure.lang.Compiler.eval(Compiler.java:6890)
	at clojure.core$eval.invokeStatic(core.clj:3105)
	at clojure.core$eval.invoke(core.clj:3101)
	at clojure.main$repl$read_eval_print__7408$fn__7411.invoke(main.clj:240)
	at clojure.main$repl$read_eval_print__7408.invoke(main.clj:240)
	at clojure.main$repl$fn__7417.invoke(main.clj:258)
	at clojure.main$repl.invokeStatic(main.clj:258)
	at clojure.main$repl.doInvoke(main.clj:174)
	at clojure.lang.RestFn.invoke(RestFn.java:1523)
	at clojure.tools.nrepl.middleware.interruptible_eval$evaluate$fn__663.invoke(interruptible_eval.clj:87)
	at clojure.lang.AFn.applyToHelper(AFn.java:152)
	at clojure.lang.AFn.applyTo(AFn.java:144)
	at clojure.core$apply.invokeStatic(core.clj:646)
	at clojure.core$with_bindings_STAR_.invokeStatic(core.clj:1881)
	at clojure.core$with_bindings_STAR_.doInvoke(core.clj:1881)
	at clojure.lang.RestFn.invoke(RestFn.java:425)
	at clojure.tools.nrepl.middleware.interruptible_eval$evaluate.invokeStatic(interruptible_eval.clj:85)
	at clojure.tools.nrepl.middleware.interruptible_eval$evaluate.invoke(interruptible_eval.clj:55)
	at clojure.tools.nrepl.middleware.interruptible_eval$interruptible_eval$fn__708$fn__711.invoke(interruptible_eval.clj:222)
	at clojure.tools.nrepl.middleware.interruptible_eval$run_next$fn__703.invoke(interruptible_eval.clj:190)
	at clojure.lang.AFn.run(AFn.java:22)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

Likewise, this produces a similar ClassCastException:

(into {} [#{:a 1}])
;=> ClassCastException ....

There doesn't seem to be any documentation on into that implies it only works when kv pairs are represented as vectors (or somehow, maps), so this seems to be a bug. It's extremely surprising that it doesn't work for pairs represented as lists.

For the interested, I found this by writing a function to invert a map in the most natural way I could think of:

(defn invert-map [m]
  (into {} (map (fn [[k v]] [v k]) m)))

(invert-map {:a 1 :b 2})
;=> {1 :a 2 :b}, no alarms and no surprises

; wait, this is pretty stupid, why don't I just use reverse...

(defn invert-map [m]
  (into {} (map reverse m)))

(invert-map {:a 1 :b 2})
;=> :(

Confirmed with Clojure 1.7 on Ubuntu 3.13.0-63-generic 64bit.

Comment by Sean Corfield [ 05/Jun/16 12:03 AM ]

There are definitely some odd edge cases around MapEntry but I would invert a map like this rather than trying to rely on a sequence of MapEntry objects:

(reduce-kv (fn [m k v] (assoc m v k)) {} m)

The fact that a map behaves as a sequence of pairs seems to cause a lot of confusion.

Comment by Alex Miller [ 05/Jun/16 3:05 PM ]

into takes a collection of elements to be conj'ed into the target collection. The differences in your examples are all around what one of those elements is, so this is really a question about conj'ing into a map. Map conj is (lightly) documented at http://clojure.org/reference/data_structures#Maps to take one of:

  • a map whose entries will be added
  • a map entry
  • a vector of 2 items

The examples you mention are lists and sets, which are none of the above. Lists are not supported because the key and value are plucked in constant time where as lists would have to be traversed sequentially to get to the 0th and 1st element. I do not think the time difference is significant but that is the philosophical reason. Sets are not supported because they are not ordered, so that to me makes no sense at all as there is no meaning of the 0th and 1st element at all.

For map-invert, you might try the one that is (obscurely) in clojure.set:

I don't see any bug here - everything is happening as designed, so I'm going to close this ticket.

Generated at Thu Apr 25 00:58:11 CDT 2019 using JIRA 4.4#649-r158309.