Incomplete (Clojure JIRA)

[CLJ-1125] Clojure can leak memory when used in a servlet container

Fri, 24 May 2013 11:36:08 -0500

When used within a servlet container
(Jetty/Tomcat/JBossAS/Immutant/etc), the thread locals Var.dvals (used
to store dynamic bindings) and LockingTransaction.transaction (used to
store the currently active transaction(s)) prevent all of the classes
loaded by an application's clojure runtime from being garbage collected,
resulting in a memory leak.

The issue comes from threads living beyond the lifetime of a
deployment - servlet containers use thread pools that are shared
across all applications within the container. Currently, the dvals and
transaction thread locals are not discarded when they are no longer
needed, causing their contents to retain a hard reference to their
classloaders, which, in turn, causes all of the classes loaded under
the application's classloader to be retained until the thread exits
(which is generally at JVM shutdown).

I've attached a patch that does the following:

Var.dvals is now removed when the thread bindings are popped
Var.dvals no longer has an initialValue, so checking to see if it is
set will no longer set it to an empty Frame
The outer transaction in LockingTransaction.transaction now removes
the thread local when it is finished

There is still the opportunity for memory leaks if agents or futures
are used, and the executors used for them are not shutdown when the
app is undeployed. That's a solvable problem, but should probably be
solved by the containers themselves (and/or the war generation tools)
instead of in clojure itself.

This patch has a small performance impact: its use of a try/finally
around running transactions to remove the outer transaction adds
4-6 microseconds to each transaction call on my hardware.

Providing an automated test for this patch is difficult - I've tested
it locally with repeated deployments to a container while monitoring
GC and permgen. All of clojure's tests pass with it applied.

The above is a condensation of:
https://groups.google.com/d/topic/clojure-dev/3CXDe8_9G58/discussion

I'm happy to provide whatever feedback/work is needed to get this
applied.

[CLJ-1083] Incorrect ArityException message for function names containing ->

Fri, 24 May 2013 11:35:12 -0500

user=> (defn a->b [])
#'user/a->b
user=> (a->b 1)
ArityException Wrong number of args (1) passed to: user$a clojure.lang.AFn.throwArity (AFn.java:437)

Note that the reported function name in the stack trace is "user$a", where it should be "user$a->b" (or some mangled variant thereof?)

See discussion here: https://groups.google.com/d/msg/clojure/PVNoLclhhB0/_NWqyE0cPAUJ

[CLJ-980] Documentation for extend-type falsely implies that & is allowed in protocol fn signatures

Wed, 19 Dec 2012 08:00:13 -0600

The documentation for extend-type contains the following example:

(extend-type MyType 
    Countable
      (cnt [c] ...)
    Foo
      (bar [x y] ...)
      (baz ([x] ...) ([x y & zs] ...)))

However, [x y & zs] is not a valid parameter list for a protocol fn. The documentation should be appropriately amended.

[CLJ-939] Exceptions thrown in the top level ns form are reported without file or line number

Sat, 15 Dec 2012 07:50:03 -0600

If there is an error in the `ns` form, an exception is thrown, which is not caught in `load`.

For example, with an invalid :only clause;

(ns clj14.myns
  (:use
   [clojure.core :only seq]))

This generates a Don't know how to create ISeq from: clojure.lang.Symbol exception, with source file or line number.

[CLJ-866] Provide a clojure.test function to run a single test case with fixtures

Sat, 4 May 2013 09:36:20 -0500

At present, clojure.test test cases are functions and can be invoked directly. However, in the case that the test relies on fixtures, this does not work. Please provide a function that can run a single test case with all fixtures applied.

[CLJ-835] defmulti doc string doesn't mention needing to be passed a var for the value of :hierarchy

Wed, 19 Dec 2012 07:59:17 -0600

The :hierarchy option for defmulti should be passed a var, but this is not mentioned in the doc string.

The error message from passing the hierarchy directly is rather cryptic:

Evaluation aborted on java.lang.ClassCastException: clojure.lang.PersistentArrayMap cannot be cast to clojure.lang.IRef (SOURCE_FORM_44:19).

[CLJ-771] Move unchecked-prim casts to clojure.unchecked

Wed, 13 Feb 2013 00:51:11 -0600

Per Rich's comment in ~~CLJ-767~~:

Moving unchecked coercions into unchecked ns is ok

[CLJ-766] Implicit casting behaviour of into-array differs from -array

Fri, 3 May 2013 20:46:29 -0500

Current patch: byte-short-array-ctors.diff

The behavior of byte-array and short-array is inconsistent with the other <type>-array functions and with the into-array function when invoked with types other than byte or short. All of the other cases upcast to Number, then extract the primitive value, allowing this operation to succeed (assuming the value is in range). byte-array and short-array throw a ClassCastException.

Example:

base64v3a=> (into-array Byte/TYPE [1 2 3 4])  ;; int to byte ok
#<byte[] [B@5ee04fd>
base64v3a=> (byte-array [1 2 3 4])  ;; int to byte NOT ok!! 
ClassCastException java.lang.Long cannot be cast to java.lang.Byte
  clojure.lang.Numbers.byte_array (Numbers.java:1418)
base64v3a=> (long-array [1 2 3 4])  ;; int to long ok
#<long[] [J@3f9f4d1d>

into-array (via RT.seqToTypedArray) and the other <type>-array functions all upcast to Number (via Numbers.<type>_array}}), then obtain the proper primitive value. Numbers.byte-array and Numbers.short-array do casts directly to Byte and Short (yielding the ClassCastException).

The attached patch makes the Byte and Short cases match the other types and the into-array behavior. Tests are included. The submitter is a contributor.

[CLJ-445] Method/Constructor resolution does not factor in widening conversion of primitive args

Mon, 20 Feb 2012 14:04:42 -0600

Problem:
When making java calls (or inlined functions), if both args and param are primitive, no widening conversion is used to locate the proper overloaded method/constructor.

Examples:

user=> (Integer. (byte 0))
java.lang.IllegalArgumentException: No matching ctor found for class java.lang.Integer (NO_SOURCE_FILE:0)
</code></pre>
The above occurs because there is no Integer(byte) constructor, though it should match on Integer(int).
<pre><code>user=> (bit-shift-left (byte 1) 1)
Reflection warning, NO_SOURCE_PATH:3 - call to shiftLeft can't be resolved.
2

In the above, a call is made via reflection to Numbers.shiftLeft(Object, Object) and its associated auto-boxing, instead of directly to the perfectly adequate Numbers.shiftLeft(long, int).

Workarounds:
Explicitly casting to the formal type.

Ancillary benefits of fixing:
It would also reduce the amount of method overloading, e.g., RT.intCast(char), intCast(byte), intCast(short), could all be removed, since such calls would pass to RT.intCast(int).

[CLJ-415] smarter assert (prints locals)

Sun, 18 Nov 2012 01:06:07 -0600

Here is an implementation you can paste into a repl. Feedback wanted:

(defn ^{:private true} local-bindings
  "Produces a map of the names of local bindings to their values."
  [env]
  (let [symbols (map key env)]
    (zipmap (map (fn [sym] `(quote ~sym)) symbols) symbols)))

(defmacro assert
  "Evaluates expr and throws an exception if it does not evaluate to
 logical true."
  {:added "1.0"}
  [x]
  (when *assert*
    (let [bindings (local-bindings &env)]
      `(when-not ~x
         (let [sep# (System/getProperty "line.separator")]
           (throw (AssertionError. (apply str "Assert failed: " (pr-str '~x) sep#
                                          (map (fn [[k# v#]] (str "\t" k# " : " v# sep#)) ~bindings)))))))))

[CLJ-373] update-in with empty key paths

Sat, 8 Sep 2012 06:13:29 -0500

To the topic of get-in and update-in. While I realize this is not a bug it is odd and in my eyes unexpected and unwanted behavior.

get-in called with an empty path returns the hashmap it was given to walk through - this is as I expect and makes a lot of sense when dynamically (with generated pathes ) walking a hash map.

update-in behaves differently and while from the implementation side, it's behavior too makes sense, it does not work as expected (at least not for me) update-in with an empty map creates a new key 'nil' so:

(update-in {...} [] f) ist he same as (update-in {...} [nil] f) while (get-in {...} []) is not the same as (get-in {...} [nil]) and of cause differs from what update-in does.

For automatically walking trees the behavior of get-in makes a lot more sense since the current behavior of update-in forces you to check for empty paths and if they are empty fall back to plain assoc and get (or get-in since this works):

(if-let [r (butlast @path)]
(do
  (alter m update-in r dissoc (last @path))
  (alter m update-in r assoc {:name @sr} c))
(do
  (alter m dissoc (last @path))
  (alter m assoc {:name @sr} c)))

Next argument is that update-in with an empty map working on nil isn't easy to gasp, one needs to know the implementation details to realize that it works, I think 90% of the people reading update-in with [] will not instinctively know that it works on the key nil, so changing this would most likely not break any current code, and if it would the code would be bad anyway .

Chouser has, a very nice solution on the mailing list that would fix the problem I'm not sure if I'm entitled to post it here since I did not wrote it but it can be found in this thread: http://groups.google.com/group/clojure/browse_thread/thread/de5b20b8c3fe498b?hl=en

Regards,
Heinz

[CLJ-346] (pprint-newline :fill) is not handled correctly

Mon, 29 Nov 2010 20:48:05 -0600

Filled pretty printing (where we try to fit as many elements on a line as possible) is being too aggressive as we can see when we try to print the following array:

user> (binding [*print-right-margin* 20] (pprint (int-array (range 10))))

Produces:

[0,
1,
2,
3,
4, 5, 6, 7, 8, 9]

Rather than

[0, 1, 2, 3, 4,
5, 6, 7, 8, 9]

or something like that. (I haven't worked through the exact correct representation for this case).

We currently only use :fill style newlines for native java arrays.

[CLJ-322] Enhance AOT compilation process to emit classfiles only for explicitly-specified namespaces

Fri, 24 May 2013 13:25:25 -0500

Summary: still needs decision on implementation approach.

This was originally/erroneously reported by Howard Lewis Ship in the clojure-contrib assembla:

My build file specifies the namespaces to AOT compile but if I include another namespace
(even from a JAR dependency) that is not AOT compiled, the other namespace will be compiled as well.

In my case, I was using clojure-contrib's clojure.contrib.str-utils2 namespace, and I got a bunch of
clojure/contrib/str_utils2 classes in my output directory.

I think that the AOT compiler should NOT precompile any namespaces that are transitively reached,
only namespaces in the set specified by the command line are appropriate.

As currently coded, you will frequently find unwanted third-party dependencies in your output JARs;
further, if multiple parties depend on the same JARs, this could cause bloating and duplication in the
eventual runtime classpath.

Having the option of shipping either all AOT-compiled classfiles or mixed source/AOT depending upon one's distribution requirements would make that phase of work with a clojure codebase significantly easier and less error-prone. The only question in my mind is what the default should be. We're all used to the current behaviour, but I'd guess that any nontrivial project where the form of the distributable matters (i.e. the source/AOT mix), providing as much control as possible by default makes the most sense. Given the tooling that most people are using, it's trivial (and common practice, IIUC) to provide a comprehensive list of namespaces one wishes to compile, so making that the default shouldn't be a hurdle to anyone. If an escape hatch is desired, a --transitive switch to clojure.lang.Compile could be added.

[CLJ-291] (take-nth 0 coll) redux...

Fri, 10 Dec 2010 08:50:19 -0600

I dont seem to be able make the old ticket uninvalid so here goes
(take-nth 0 coll) causes (at least on Solaris) infinite space and time consumption
It's not a printout error as the following code causes the problem too

(let [j 0
firstprod (apply * (doall (map #(- 1 %) (take-nth j (:props mix)))))]) ; from my parameter update function

I used jvisualvm and the jvm is doing some RNI call - no clojure code is running at all
If left alone it will crash the jvm with all heap space consumed
0 is an InvalidArgument for take-nth
I wouldnt mind if it produced an infinite lazy sequence of nils even though thats wrong
It doesnt do this though it actively destroys the JVM
Its a bug nasty destructive and it took me half a day to figure out what was going on
please let someone fix it!

[CLJ-250] debug builds

Thu, 14 Mar 2013 13:28:51 -0500

This ticket includes two patches:

a patch to set assert when clojure.lang.RT loads, based on the presence of system property clojure.debug
expand error messages in assert to include local-bindings</code> (a new macro which wraps the implicit <code>&env)

Things to consider before approving these patches:

should there be an easy Clojure-level way to query if debug is enabled? (checking assert isn't the same, as debug should eventually drive other features)
assertions will now be off by default – this is a change!
is the addition of the name local-bindings to clojure.core cool?

[CLJ-211] Support arbitrary functional destructuring via -> and ->>

Fri, 10 Dec 2010 08:49:03 -0600

Support arbitrary functional destructuring, that is the use of
any function in any destructuring form to help unpack data in
arbitrary ways.

The discussion began here:
http://clojure-log.n01se.net/date/2009-11-17.html#09:31c

The attached patch implements the spec described here:
http://clojure-log.n01se.net/date/2009-11-17.html#10:50a

That is, the following examples would now work:

user=> (let [(-> str a) 1] a)
"1"

user=> (let [[a (-> str b) c] [1 2]] (list a b c))
(1 "2" nil)

user=> (let [(->> (map int) [a b]) "ab"] (list a b))
(97 98)

[CLJ-84] GC Issue 81: compile gen-class fail when class returns self

Fri, 10 Dec 2010 09:30:52 -0600

Reported by davidhaub, Feb 14, 2009

When attempting to compile the following program, clojure fails with a
ClassNotFoundException.  It occurs because one of the methods returns the
same class that is being generated.  If the returnMe method below is
changed to return an Object, the compile succeeds.

Beware when testing! If the classpath contains a class file (say from a
prior successful build when the returnMe method was changed to return an
object), the compile will succeed.  Always clear out the
clojure.compile.path prior to compiling.

;badgenclass.clj
(ns badgenclass
  (:gen-class
     :state state
     :methods
     [[returnMe [] badgenclass]]
     :init init))
(defn -init []
  [[] nil])

(defn -returnMe [this]
  this)

#!/bin/sh
rm -rf classes
mkdir classes
java -cp lib/clojure.jar:classes:. -Dclojure.compile.path=classes \
clojure.lang.Compile badgenclass


Comment 1 by chouser, Mar 07, 2009

Attached is a patch that accepts strings or symbols for parameter and return class
names, and generates the appropriate bytecode without calling Class/forName.  It
fixes this issue, but because 'ns' doesn't resolve :gen-class's arguments, class
names aren't checked as early anymore.  :gen-class-created classes with invalid
parameter or return types can even be instantiated, and no error will be reported
until the broken method is called.

One possible alternative would be to call Class/forName on any symbols given, but
allow strings to use the method given by this patch.  To return your own type, you'd
need a method defined like:

  [returnMe [] "badgenclass"]

Any thoughts?

[CLJ-69] GC Issue 66: Add "keyset" to Clojure; make .keySet for APersistentMap return IPersistentSet

Fri, 10 Dec 2010 08:48:40 -0600

Reported by wolfe.a.jason, Feb 04, 2009

Describe the feature/change.

Add "keyset" to Clojure; make .keySet for APersistentMap return an
IPersistentSet

Was this discussed on the group? If so, please provide a link to the
discussion:

http://groups.google.com/group/clojure/browse_thread/thread/66e708e477ae992f/ff3d8d588068b60e?hl=en#ff3d8d588068b60e

-----------------------------------------------------

A patch is attached.  Some notes:

I chose to add a "keyset" function, rather than change the existing "keys",
so as to avoid breaking anything.

The corresponding RT.keyset function just calls .keySet on the argument.
I would have liked to have "keyset" return an IPersistentSet even when
passed a (non-Clojure) java.util.Map, but this seems impossible to do in
sublinear time because of essentially the same limitation mentioned in the
above thread (the Map interface does not support getKey() or entryAt()) --
assuming, again, that "get" is supposed to return the actual (identical?)
key in a set, and not just an .equal key.

I then changed the implementation of .keySet for APersistentMap to
essentially copy APersistentSet.  A more concise alternative would have
been to extend APersistentSet and override the .get method, but that made
me a bit nervous (since if APeristentSet changed this could break). 

Anyway, this is my first patch for the Java side of Clojure, and I'm not
yet solid on the conventions and aesthetics, so
comments/questions/criticisms/requests for revisions are very welcome.

[CLJ-5] Unintuitive error response in clojure 1.0

Sat, 28 Apr 2012 22:48:40 -0500

The following broken code:

(let [[x y] {}] x)

provides the following stack trace:

Exception in thread "main" java.lang.UnsupportedOperationException: nth not supported on this type: PersistentArrayMap (test.clj:0)
at clojure.lang.Compiler.eval(Compiler.java:4543)
at clojure.lang.Compiler.load(Compiler.java:4857)
at clojure.lang.Compiler.loadFile(Compiler.java:4824)
at clojure.main$load_script__5833.invoke(main.clj:206)
at clojure.main$script_opt__5864.invoke(main.clj:258)
at clojure.main$main__5888.doInvoke(main.clj:333)
at clojure.lang.RestFn.invoke(RestFn.java:413)
at clojure.lang.Var.invoke(Var.java:346)
at clojure.lang.AFn.applyToHelper(AFn.java:173)
at clojure.lang.Var.applyTo(Var.java:463)
at clojure.main.main(main.java:39)
Caused by: java.lang.UnsupportedOperationException: nth not supported on this type: PersistentArrayMap
at clojure.lang.RT.nth(RT.java:800)
at clojure.core$nth__3578.invoke(core.clj:873)
at user$eval__1.invoke(test.clj:1)
at clojure.lang.Compiler.eval(Compiler.java:4532)
... 10 more

The message "nth not supported on this type" while correct doesn't make the cause of the error very clear. Better error messages when destructuring would be very helpful.