This repository provides experiment results of ConFix applied to Defejects4j dataset.

For detailed approaches and evaluation results, you may check ConFix manuscript.

For ConFix source code, please check here.
Note that code can be updated from the version we used for experiments.

patches directory contains Human and ConFix patches.
analysis contains information related to ConFix's strategy analysis.
coverage.tar.gz contains coverage-info.obj files for Defects4j bugs used for experiments.
analysis contains information related to ConFix's strategy analysis.

For replication, we provide a binary and a sample in replication directory.

pool contains PTLRH and PLRT change pools used for experiments.
Files are splitted due to GitHub's file size limitation.
You can use following command to unarchive files.
cat pool.tar.gz.* | tar xz

All coverage information, contexts and changes are stored in .obj files, which you may find after unarchiving coverage.tar.gz or pool.tar.gz.*.
They are object instances stored by ObjectOutputStream. For example, coverage-info.obj can be read as a com.github.thwak.confix.coverage.CoverageManager class instance.

Normally, these files are automatically read and used by ConFix. However, these information can be also accessed outside ConFix by following methods.

As mentioned above, you can simply use ObjectInputStream to read coverage-info.obj files as CoverageManager class instance.
In your Java code, import com.github.thwak.confix.coverage.CoverageManager class which can be obtained from ConFix implmentation, and read .obj file as an instance.
To obtained a list of covered lines, you can use computeScore() method like the following.

CoverageManager manager;
//Read .obj file
List<CoveredLine> lines = manager.computeScore("ochiai");

This will return a list sorted in descending order of FL scores computed by Ochiai, and each CoveredLine in the returned list contains class name, line number, and computed FL score.

For more supported functionalities, please refer to the actual implementation.

Change pools have more complex structures.
We recommend to instantiate com.github.thwak.confix.pool.ChangePool with a specific poolDir like the following.

ChangePool pool = new ChangePool(new File("pool/ptlrh"));

Above code create a change pool attached to a pool directory pool/ptlrh.
Note that you can obtain PTLRH and PLRT changes pools from pool.tar.gz.* files.

To read stored information - contexts, changes, etc. - from a pool directory, you can simply call pool.load().
Then you can access to collected contexts and changes by using getContexts() or getChange() methods.
For details, please check ConFix implementation.
There are other methods which ChangePool supports to utilize collected changes, such as getting change frequencies or retrieving changes with a certain context.

patches directory contains Human and ConFix patches. Directories and files in patches have the following structure.

• project
  • bugname
    • buggy (Buggy code)
    • confix (ConFix patch)
      • edit (The applied change to generate ConFix patch)
    • human (Human-written patch)
    • confix.diff (Differences between buggy code & ConFix patch.)
    • human.diff (Differences between buggy code & human-written patch.)
    • patch_info (Simple info. on ConFix patch)

Each edit file contains detailed information of the applied change like the following.

Class:org.jfree.chart.block.BorderArrangement //modified class
update
w
to
h
at line 450 //string representation of the change.
Applied Change
update
var0
at ArrayAccess::null[array]
to
var0 //applied abstract change in change pool
from
hadoop-0815a6f:hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/web/resources/NamenodeWebHdfsMethods.java: //where the change was collected
19 //frequency in change pool
P:ArrayAccess{array},L:,R:{34} //change's context