Source code vs. Docker image

[jose]

Hi,

(This is a follow up of issue #19).

The execution of, for example, Nopol on Closure-51 produces a different outcome when executed from the source code or via the provided docker image.

Nopol generates the following patch in a few minutes when executed from RepairThemAll's source code:

+++ b/src/com/google/javascript/rhino/Node.java
@@ -1566,7 +1566,9 @@
       if (diff != null) {
-        return "Node tree inequality:" +
-            "\nTree1:\n" + toStringTree() +
-            "\n\nTree2:\n" + node2.toStringTree() +
-            "\n\nSubtree1: " + diff.nodeA.toStringTree() +
-            "\n\nSubtree2: " + diff.nodeB.toStringTree();
+        if ((node2.type) != (com.google.javascript.rhino.Node.this.type)) {
+          return "Node tree inequality:" +
+          "\nTree1:\n" + toStringTree() +
+          "\n\nTree2:\n" + node2.toStringTree() +
+          "\n\nSubtree1: " + diff.nodeA.toStringTree() +
+          "\n\nSubtree2: " + diff.nodeB.toStringTree();
+        }
       }

However, when Nopol is executed via the provided docker image it crashes:

...
18:14:34.750 [main] INFO  fr.inria.lille.repair.nopol.NoPol - Java version: 1.8.0_111
18:14:34.750 [main] INFO  fr.inria.lille.repair.nopol.NoPol - JAVA_HOME: /usr/lib/jvm/java-1.8.0-openjdk-amd64/bin/
18:14:34.751 [main] INFO  fr.inria.lille.repair.nopol.NoPol - PATH: /usr/lib/jvm/java-1.8.0-openjdk-amd64/bin/:/defects4j/framework/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
java.util.concurrent.ExecutionException: java.lang.NullPointerException
	at java.util.concurrent.FutureTask.report(FutureTask.java:122)
	at java.util.concurrent.FutureTask.get(FutureTask.java:206)
	at fr.inria.lille.repair.Main.main(Main.java:106)
Caused by: java.lang.NullPointerException
	at com.gzoltar.core.GZoltar.run(GZoltar.java:51)
	at fr.inria.lille.localization.GZoltarFaultLocalizer.run(GZoltarFaultLocalizer.java:163)
	at fr.inria.lille.localization.GZoltarFaultLocalizer.<init>(GZoltarFaultLocalizer.java:94)
	at fr.inria.lille.localization.GZoltarFaultLocalizer.<init>(GZoltarFaultLocalizer.java:68)
	at fr.inria.lille.localization.GZoltarFaultLocalizer.createInstance(GZoltarFaultLocalizer.java:60)
	at fr.inria.lille.repair.nopol.NoPol.createLocalizer(NoPol.java:179)
	at fr.inria.lille.repair.nopol.NoPol.build(NoPol.java:131)
	at fr.inria.lille.repair.Main$1.call(Main.java:101)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

Usage: java -jar nopol.jar
                          [(-m|--mode) <repair|ranking>] (-e|--type) <condition|precondition|pre_then_cond|loop|arithmetic> [(-o|--oracle) <angelic|symbolic>] [(-y|--synthesis) <smt|dynamoth>] [(-l|--solver) <z3|cvc4>] [(-p|--solver-path) <solverPath>] (-s|--source) source1:source2:...:sourceN  (-c|--classpath) <classpath> [(-t|--test) test1:test2:...:testN ] [--complianceLevel <complianceLevel>] [--maxTime <maxTime>] [--maxTimeType <maxTimeType>] [(-z|--flocal) < cocospoon|dumb|gzoltar>] [--output <outputFolder>] [--json[:<outputJson>]]
...

(here is the repair.log file)

The problem does not seem to be due to different classpaths (more info in here), and neither due to a different version of Java 8 (more info in here). The provided docker image might be (somehow) using a different version of Nopol... 🤷🏻‍♂️

--
Best,
Jose

[tdurieux]

It is not a different version of nopol, we never changed it during the experiment and the docker image is really based on the content of this repo.

The error in the Docker image is triggered because no failing test case has been found thus the fault localization is failing.

Are you using oracle jdk locally or openjdk?

[jose]

Are you using oracle jdk locally or openjdk?

I've tried Oracle JDK and OpenJDK in the Docker image and got the same NullPointerException.