Using Fuzzy Code Search to Link Code Fragments in Discussions to Source Code

Using Fuzzy Code Search to Link
Code Fragments in Discussions to
Source Code

Nicolas Bettenburg Stephen W. Thomas Ahmed E. Hassan
Queen’s University
Software Analysis and Intelligence Lab (SAIL)
Kingston, Ontario, Canada

“Given enough eyeballs,
all bugs are shallow.”
Eric S. Raymond - The Cathedral and the Bazaar

2

Objective in this Study:

Which parts of the
source code did
developers
discuss?

3

Objective in this Study:

Which parts of the
source code did
developers
discuss?
Finding Associations of Discussions with Source Code Files

3

Communication Source Code
Repository Archive
4

Untitled 12-03-15 2:55 PM


this.opt_withObsolete = opt_withObsolete;
Untitled
} 12-03-15 2:55 PM

/* (non-Javadoc)
* @see main.IExperiment#getDescription()
}
*/
public String getDescription() {
/* (non-Javadoc)
String description = "We want to run an experiment all those bug
reports.nt"
}
*/
+ "We extract the complete stacktrace information to study fix
location and stacktrace interplay.nt";
/* (non-Javadoc)
return description;
reports.nt"
}
} */
/* (non-Javadoc)
/* (non-Javadoc) String description = "We want to run an experiment all those bug
return description;
* @see main.IExperiment#run()
reports.nt"
} */
*/ + "We extract the complete stacktrace information to study fix
public boolean run() { location and stacktrace interplay.nt";
final boolean return = Main.isCli_withDebugMessages();
DEBUG description;
* @see main.IExperiment#run()reports.nt"
}
// Select an appropriate database name on the server depending on the
project demanded via command line options
/* (non-Javadoc)
DEBUG description;
String dbName = "";
}
if (Main.getCli_project().equalsIgnoreCase("eclipse"))
*/
dbName = "eclipse-december-2007";
public boolean run() {
/* (non-Javadoc)
else { final boolean DEBUG = Main.isCli_withDebugMessages();
String dbName = "";
System.out.println("We currently only support the Eclipse database
*/
!"); // Select an appropriate database name on the server depending on the
System.exit(0); project demanded via command line options
} String dbName = "";
// Create the else {Database Connection
DatabaseConnector System.out.println("We currently only support the Eclipse database
dbc = new DatabaseConnector(Main.getCli_dbHost(),
Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
!");
dbc.initialize(); System.exit(0);
// Create the else { Database Connection
if (!dbc.connect()) {
}
System.out.println("Could not connect to database. Check host, port
!");
, username and password.");
System.exit(1);
// Create the Database Connection
}
} DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
// Fetch a number of Bug Report unique IDs belonging to the following
dbc.initialize();
System.exit(1);
query if (!dbc.connect()) {
String setSelectionQuery = "SELECT DISTINCT not connect to database. Check host, port
System.out.println("Could bug_id FROM bugzilla_bugs
WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
dbc.initialize();
LIMIT " + Main.getCli_amount();
System.exit(1);
int[] bugIDs = dbc.getIdSetByQuery(setSelectionQuery);
}
int lastID = bugIDs[bugIDs.length -1];
System.exit(1);
if (DEBUG) query
}
System.out.println("* Attempting to fetch " + bugIDs.length + " bugzilla_bugs
String setSelectionQuery = "SELECT DISTINCT bug_id FROM
reports from " + Main.getCli_dbHost() + " starting +with ID " BY bug_id ASC
WHERE bug_id > " + Main.getCli_startID() " ORDER +
Main.getCli_startID());
if (DEBUG) query
// Collect the bug reports bugIDs[bugIDs.length -1];up to 100 parallel
int lastID = from the database using
threads.
if (DEBUG)
System.out.println("* Attempting to fetch " + bugIDs.length + "
reports from " + Main.getCli_dbHost() + " starting with ID " +
threads.
if (DEBUG)
// Collect the bug reports from the database using up to 100Page 2 of 4
parallel
threads.

parallel
threads.

Page 2 of 4

Page 2 of 4

Repository Archive
4



Untitled
} 12-03-15 2:55 PM

/* (non-Javadoc)
}
*/
/* (non-Javadoc)
reports.nt"
}
*/
/* (non-Javadoc)
return description;
reports.nt"
}
} */
/* (non-Javadoc)
return description;
reports.nt"
} */
DEBUG description;
* @see main.IExperiment#run()reports.nt"
}
/* (non-Javadoc)

Traceability
DEBUG description;
String dbName = "";
}
*/
/* (non-Javadoc)
String dbName = "";
*/
// Create the else {Database Connection
!");
// Create the else { Database Connection
}
!");

Links
System.exit(1);
}
dbc.initialize();
System.exit(1);
dbc.initialize();
System.exit(1);
}
System.exit(1);
if (DEBUG) query
}
if (DEBUG) query
threads.
if (DEBUG)
threads.
if (DEBUG)
parallel
threads.

parallel
threads.

Page 2 of 4

Page 2 of 4

Repository Archive
4

org.eclipse.swt/Eclipse_SWT/win32/org/eclipse/swt/widgets/Button.java

5

3 Approaches in the Past

Change Log Analysis

Information Retrieval

Lightweight Textual Analysis

6

Change Log Analysis


}

/* (non-Javadoc)
*/
reports.nt"
return description;
}

/* (non-Javadoc)

LUCENE-2297: allow reader
*/
final boolean DEBUG = Main.isCli_withDebugMessages();

Untitled 12-03-15 2:55project demanded via command line options
PM
String dbName = "";
else {
!");
this.opt_withObsolete = opt_withObsolete; System.exit(0);
} }
/* (non-Javadoc)
* @see main.IExperiment#getDescription() // Create the Database Connection
*/ DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
public String getDescription() { Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
String description = "We want to run an experiment all those bug dbc.initialize();
reports.nt" if (!dbc.connect()) {

pooling to be enabled (through
+ "We extract the complete stacktrace information to study fix System.out.println("Could not connect to database. Check host, port
location and stacktrace interplay.nt"; , username and password.");
return description; System.exit(1);
} }
/* (non-Javadoc) // Fetch a number of Bug Report unique IDs belonging to the following
* @see main.IExperiment#run() query
*/ String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs
public boolean run() { WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
final boolean DEBUG = Main.isCli_withDebugMessages(); LIMIT " + Main.getCli_amount();
// Select an appropriate database name on the server depending on the int lastID = bugIDs[bugIDs.length -1];
String dbName = ""; if (DEBUG)
if (Main.getCli_project().equalsIgnoreCase("eclipse")) System.out.println("* Attempting to fetch " + bugIDs.length + "
dbName = "eclipse-december-2007"; reports from " + Main.getCli_dbHost() + " starting with ID " +
else { Main.getCli_startID());
!"); // Collect the bug reports from the database using up to 100 parallel
System.exit(0); threads.
}

IndexWriterConfig) even if
Untitled // Create the DatabasePM
12-03-15 2:55 Connection
Untitled
DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(), 12-03-15 2:55 PM
dbc.initialize(); Page 2 of 4
this.opt_withObsolete = opt_withObsolete; , username and password.");
} System.exit(1); this.opt_withObsolete = opt_withObsolete;
} }
/* (non-Javadoc)
* @see main.IExperiment#getDescription() // Fetch a number of Bug Report unique IDs belonging to the following
/* (non-Javadoc)
*/ query * @see main.IExperiment#getDescription()
public String getDescription() { String setSelectionQuery = "SELECT DISTINCT bug_id FROM */bugzilla_bugs
WHERE bug_id > " + Main.getCli_startID() + " ORDER public String getDescription() {
String description = "We want to run an experiment all those bug BY bug_id ASC
reports.nt" LIMIT " + Main.getCli_amount(); String description = "We want to run an experiment all those bug
+ "We extract the complete stacktrace information to study fix reports.nt"
location and stacktrace interplay.nt"; int lastID = bugIDs[bugIDs.length -1]; + "We extract the complete stacktrace information to study fix
return description; location and stacktrace interplay.nt";
if (DEBUG)

you're not pulling NRT reader.
} return description;
}
/* (non-Javadoc) reports from " + Main.getCli_dbHost() + " starting with ID " +
* @see main.IExperiment#run() Main.getCli_startID()); /* (non-Javadoc)
*/ * @see main.IExperiment#run()
public boolean run() { // Collect the bug reports from the database using up to 100 parallel
*/
final boolean DEBUG = Main.isCli_withDebugMessages(); threads. public boolean run() {
project demanded via command line options // Select an appropriate database name on the server depending on the
String dbName = ""; project demanded via command line options
if (Main.getCli_project().equalsIgnoreCase("eclipse")) String dbName = "";
dbName = "eclipse-december-2007"; if (Main.getCli_project().equalsIgnoreCase("eclipse"))
Page 2 of 4
else { dbName = "eclipse-december-2007";
System.out.println("We currently only support the Eclipse database else {
!"); System.out.println("We currently only support the Eclipse database
System.exit(0); !");
} System.exit(0);
}

DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(), // Create the Database Connection
Main.getCli_dbUser(), Main.getCli_dbPass(), dbName); DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
dbc.initialize(); Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
if (!dbc.connect()) { dbc.initialize();
System.out.println("Could not connect to database. Check host, port if (!dbc.connect()) {
, username and password."); System.out.println("Could not connect to database. Check host, port
System.exit(1); , username and password.");
} System.exit(1);
}
query // Fetch a number of Bug Report unique IDs belonging to the following
String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs query
WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs
LIMIT " + Main.getCli_amount(); WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
int[] bugIDs = dbc.getIdSetByQuery(setSelectionQuery); LIMIT " + Main.getCli_amount();
int lastID = bugIDs[bugIDs.length -1]; int[] bugIDs = dbc.getIdSetByQuery(setSelectionQuery);
if (DEBUG)
System.out.println("* Attempting to fetch " + bugIDs.length + " if (DEBUG)
reports from " + Main.getCli_dbHost() + " starting with ID " + System.out.println("* Attempting to fetch " + bugIDs.length + "
Main.getCli_startID()); reports from " + Main.getCli_dbHost() + " starting with ID " +
// Collect the bug reports from the database using up to 100 parallel
threads. // Collect the bug reports from the database using up to 100 parallel
threads.

SVN
Page 2 of 4
Page 2 of 4

7

Change Log Analysis

Bug Report
#LUCENE-2297


}

/* (non-Javadoc)
*/
reports.nt"
return description;
}

/* (non-Javadoc)

LUCENE-2297: allow reader
*/

Untitled 12-03-15 2:55project demanded via command line options
PM
String dbName = "";
else {
!");
this.opt_withObsolete = opt_withObsolete; System.exit(0);
} }
/* (non-Javadoc)
* @see main.IExperiment#getDescription() // Create the Database Connection
*/ DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
public String getDescription() { Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
String description = "We want to run an experiment all those bug dbc.initialize();
reports.nt" if (!dbc.connect()) {

pooling to be enabled (through
+ "We extract the complete stacktrace information to study fix System.out.println("Could not connect to database. Check host, port
location and stacktrace interplay.nt"; , username and password.");
return description; System.exit(1);
} }
/* (non-Javadoc) // Fetch a number of Bug Report unique IDs belonging to the following
* @see main.IExperiment#run() query
*/ String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs
public boolean run() { WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
final boolean DEBUG = Main.isCli_withDebugMessages(); LIMIT " + Main.getCli_amount();
// Select an appropriate database name on the server depending on the int lastID = bugIDs[bugIDs.length -1];
String dbName = ""; if (DEBUG)
if (Main.getCli_project().equalsIgnoreCase("eclipse")) System.out.println("* Attempting to fetch " + bugIDs.length + "
dbName = "eclipse-december-2007"; reports from " + Main.getCli_dbHost() + " starting with ID " +
else { Main.getCli_startID());
!"); // Collect the bug reports from the database using up to 100 parallel
System.exit(0); threads.
}

IndexWriterConfig) even if
Untitled // Create the DatabasePM
12-03-15 2:55 Connection
Untitled
DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(), 12-03-15 2:55 PM
dbc.initialize(); Page 2 of 4
this.opt_withObsolete = opt_withObsolete; , username and password.");
} System.exit(1); this.opt_withObsolete = opt_withObsolete;
} }
/* (non-Javadoc)
* @see main.IExperiment#getDescription() // Fetch a number of Bug Report unique IDs belonging to the following
/* (non-Javadoc)
*/ query * @see main.IExperiment#getDescription()
public String getDescription() { String setSelectionQuery = "SELECT DISTINCT bug_id FROM */bugzilla_bugs
WHERE bug_id > " + Main.getCli_startID() + " ORDER public String getDescription() {
String description = "We want to run an experiment all those bug BY bug_id ASC
reports.nt" LIMIT " + Main.getCli_amount(); String description = "We want to run an experiment all those bug
+ "We extract the complete stacktrace information to study fix reports.nt"
location and stacktrace interplay.nt"; int lastID = bugIDs[bugIDs.length -1]; + "We extract the complete stacktrace information to study fix
return description; location and stacktrace interplay.nt";
if (DEBUG)

you're not pulling NRT reader.
} return description;
}
/* (non-Javadoc) reports from " + Main.getCli_dbHost() + " starting with ID " +
* @see main.IExperiment#run() Main.getCli_startID()); /* (non-Javadoc)
*/ * @see main.IExperiment#run()
public boolean run() { // Collect the bug reports from the database using up to 100 parallel
*/
final boolean DEBUG = Main.isCli_withDebugMessages(); threads. public boolean run() {
project demanded via command line options // Select an appropriate database name on the server depending on the
String dbName = ""; project demanded via command line options
if (Main.getCli_project().equalsIgnoreCase("eclipse")) String dbName = "";
dbName = "eclipse-december-2007"; if (Main.getCli_project().equalsIgnoreCase("eclipse"))
Page 2 of 4
else { dbName = "eclipse-december-2007";
System.out.println("We currently only support the Eclipse database else {
!"); System.out.println("We currently only support the Eclipse database
System.exit(0); !");
} System.exit(0);
}

DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(), // Create the Database Connection
Main.getCli_dbUser(), Main.getCli_dbPass(), dbName); DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
dbc.initialize(); Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
if (!dbc.connect()) { dbc.initialize();
System.out.println("Could not connect to database. Check host, port if (!dbc.connect()) {
, username and password."); System.out.println("Could not connect to database. Check host, port
System.exit(1); , username and password.");
} System.exit(1);
}
query // Fetch a number of Bug Report unique IDs belonging to the following
String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs query
WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs
LIMIT " + Main.getCli_amount(); WHERE bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
int[] bugIDs = dbc.getIdSetByQuery(setSelectionQuery); LIMIT " + Main.getCli_amount();
int lastID = bugIDs[bugIDs.length -1]; int[] bugIDs = dbc.getIdSetByQuery(setSelectionQuery);
if (DEBUG)
System.out.println("* Attempting to fetch " + bugIDs.length + " if (DEBUG)
reports from " + Main.getCli_dbHost() + " starting with ID " + System.out.println("* Attempting to fetch " + bugIDs.length + "
Main.getCli_startID()); reports from " + Main.getCli_dbHost() + " starting with ID " +
threads. // Collect the bug reports from the database using up to 100 parallel
threads.

SVN
Page 2 of 4
Page 2 of 4

7

Information Retrieval
Description Map<Integer, BugReport> bugReports
Jared Burns 2002-11-01 14:56:40 EST = dbc.getReportsConcurrent(bugIDs,
Build 20021031(, 30, and 29) Main.isCli_withDiscussions(),
Linux-GTK Main.isCli_withAttachments(),
true); // Instantiate Output
Very often (in recent builds), when I Class // This time we will need to
perform an action that requires use some XML Outputter for the
establishing a connection to the Stacktraces
dev.eclipse.org CVS server, the Experiment_Structural_Logger logger
progress dialog appears but it stalls = new Experiment_Structural_Logger
before any progress is made. I have (); DataExportUtility dexutil = new
to hit Cancel, wait 10+ seconds (I DataExportUtility(); XMLOutputter
just switch apps), and then try again outputter = new XMLOutputter
or it never (Format.getPrettyFormat()); Element
connects. rootElement = new Element
("CVS"); // Process the Reports if
This happens to me most often when I (Main.getCli_project
click the "Synchronize all CVS ().equalsIgnoreCase("eclipse"))
projects..." button. The dialog { for (int id : bugIDs) { }
appears and says "Authenticating over BugReport currentReport =
extssh", but the progress monitor bugReports.get(id); Element
never budges. reportElement = new Element
Discussion ("Report"); Code
8


I20041130-0800

Wrong compiler error when interface overrides two methods with same signature
but different thrown exceptions: The call to ij.m() is OK, but eclipse flags it
with "Unhandled exception type IOException".

public class Over {
void x() throws ZipException {
IandJ ij= new K();
ij.m(); //wrong compile error
}
void y() throws ZipException {
K k= new K();
k.m();
}
}

interface I { void m() throws IOException; }
interface J { void m() throws ZipException; }
interface IandJ extends I, J {} // swap I and J to make compile error disappear
class K implements IandJ { public void m() throws ZipException { } }

9


ZipExcept
K.java Over.java ion.java I.java J.java

I20041130-0800

Wrong compiler error when interface overrides two methods with same signature IOExcepti
but different thrown exceptions: The call to ij.m() is OK, but eclipse flags it on.java

public class Over {
IandJ ij= new K();
}
K k= new K();
k.m();
}
}


9

X Change Log Analysis
Generalizeability: Only works for Bug Reports
Granularity: Associates complete discussion thread
Precision: Link to all ﬁles changed in commit
Conceptual Mismatch: Ignores the actual discussions!

10


X Information Retrieval
Generalizeability: Works for all kinds of documents
Granularity: Which granularity level to choose?
Precision: Lots and lots of false positives!
Conceptual Mismatch: Links unrelated documents!

10


X Information Retrieval
Granularity: Which granularity level to choose?
Precision: Lots and lots of false positives!
Conceptual Mismatch: Links unrelated documents!

X Lightweight Textual Analysis
Granularity: Very ﬁne-grained
Precision: Up to 95% precision [Bacchelli-ICSE10]
Conceptual Mismatch: Links to every class mentioned
10

In this Paper:
Fuzzy Code Search
ZipExcept IOExcepti
I20041130-0800 K.java Over.java ion.java I.java J.java on.java


public class Over {
IandJ ij= new K();
}
K k= new K();
k.m();
}
}


11

In this Paper:
Fuzzy Code Search
ZipExcept IOExcepti
I20041130-0800 K.java Over.java ion.java I.java J.java on.java


public class Over {
IandJ ij= new K();
ij.m(); //wrong compile error X.java
}
K k= new K();
k.m();
}
} Y.java


11

Fuzzy Code Search
Approach

Discussion

12

Fuzzy Code Search
Approach

Discussion

infoZilla tool
12

Fuzzy Code Search
Approach

Code Fragments
Discussion

infoZilla tool
12

Fuzzy Code Search

}
/* (non-Javadoc)
*/
reports.nt" this.opt_withObsolete = opt_withObsolete;
+ "We extract} the complete stacktrace information to study fix
return description; (non-Javadoc)
/*
} * @see main.IExperiment#getDescription()
*/
/* (non-Javadoc) public String getDescription() {
* @see main.IExperiment#run() description = "We want to run an experiment all those bug
String
*/ reports.nt"
public boolean run() { + "We extract the complete stacktrace information to study fix
final boolean DEBUG = Main.isCli_withDebugMessages();interplay.nt";
location and stacktrace
return description;

Clone Detection
}
String dbName = ""; (non-Javadoc)
/*
*/
else { public boolean run() {
System.out.println("We boolean DEBUG = support the Eclipse database
final currently only Main.isCli_withDebugMessages();
!");
System.exit(0); // Select an appropriate database name on the server depending on the
} project demanded via command line options
String dbName = "";
// Create the Database Connection "eclipse-december-2007";
dbName =
DatabaseConnector dbcelse { DatabaseConnector(Main.getCli_dbHost(),
= new
Main.getCli_dbUser(),System.out.println("We currently only support the Eclipse database
Main.getCli_dbPass(), dbName);
dbc.initialize(); !");
if (!dbc.connect()) { System.exit(0);
}
System.exit(1);
} // Create the Database Connection
DatabaseConnector dbc = new DatabaseConnector(Main.getCli_dbHost(),
// Fetch a number of Bug Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
Report unique IDs belonging to the following
query dbc.initialize();
String setSelectionQuery(!dbc.connect()) {
if = "SELECT DISTINCT bug_id FROM bugzilla_bugs
WHERE bug_id > " + Main.getCli_startID() + " ORDER connect to ASC
System.out.println("Could not BY bug_id database. Check host, port

Approach
System.exit(1);
}

if (DEBUG) // Fetch a number of Bug Report unique IDs belonging to the following
System.out.println("*query
Attempting to fetch " + bugIDs.length + "
reports from String setSelectionQuery = "SELECT DISTINCT ID " + FROM bugzilla_bugs
" + Main.getCli_dbHost() + " starting with bug_id
Main.getCli_startID());bug_id > " + Main.getCli_startID() + " ORDER BY bug_id ASC
WHERE
// Collect the bug reports bugIDs = dbc.getIdSetByQuery(setSelectionQuery);
int[] from the database using up to 100 parallel
threads. int lastID = bugIDs[bugIDs.length -1];

if (DEBUG)
Page 2 of 4
threads.

Page 2 of 4

}
/* (non-Javadoc)
*/
reports.nt" this.opt_withObsolete = opt_withObsolete;
+ "We extract} the complete stacktrace information to study fix
/*
*/
/* (non-Javadoc) public String getDescription() {
String
*/ reports.nt" this.opt_withObsolete = opt_withObsolete;
public boolean run() { + "We extract} the complete stacktrace information to study fix
final boolean DEBUG = Main.isCli_withDebugMessages();interplay.nt";
/*
*/
/* public String getDescription() {
String
*/ reports.nt"
else { public boolean run() { + "We extract the complete stacktrace information to study fix
System.out.println("We boolean DEBUG = support the Eclipse database
final currently only Main.isCli_withDebugMessages();interplay.nt";
!");
return description;
System.exit(0); // Select an appropriate database name on the server depending on the
}
/*
// Create the Database Connection "eclipse-december-2007";
dbName = */
= new
Main.getCli_dbUser(),System.out.println("We currently only support the Eclipse database
Main.getCli_dbPass(), dbName);
dbc.initialize(); !");
if (!dbc.connect()) { System.exit(0); // Select an appropriate database name on the server depending on the
String dbName = "";
System.exit(1);
} // Create the Database Connection "eclipse-december-2007";
dbName =
= new
// Fetch a number of Bug Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
query dbc.initialize(); !");
String setSelectionQuery(!dbc.connect()) {
if = "SELECT DISTINCT bug_id FROM bugzilla_bugs
System.exit(0);
WHERE bug_id > " + Main.getCli_startID() + " ORDER connect to ASC
System.out.println("Could not BY bug_id database. Check host, port
}
LIMIT " + Main.getCli_amount();, username and password.");
System.exit(1);
} // Create the Database Connection
if (DEBUG) // Fetch a number of Bug Main.getCli_dbUser(), Main.getCli_dbPass(), dbName);
System.out.println("*queryAttempting to fetch " + bugIDs.length + "
dbc.initialize();
WHERE System.out.println("Could not connect to database. Check host, port
System.exit(1);
}

if (DEBUG) // Fetch a number of Bug Report unique IDs belonging to the following
System.out.println("*query
Attempting to fetch " + bugIDs.length + "
WHERE
LIMIT " + Main.getCli_amount(); 2 of 4
Page

if (DEBUG)
Page 2 of 4
threads.

Page 2 of 4


}

/* (non-Javadoc)
*/
reports.nt"
return description;
}

/* (non-Javadoc)
*/

String dbName = "";
else {
!");
System.exit(0);
}

dbc.initialize();
System.exit(1);
}

query
String setSelectionQuery = "SELECT DISTINCT bug_id FROM bugzilla_bugs

if (DEBUG)

threads.

Page 2 of 4

Code Fragments
Discussion

infoZilla tool
12

Evaluation of our Approach

13


1) Quantitative
Do developers talk about the things they
end up changing?

13


1) Quantitative
Do developers talk about the things they
end up changing?

2) Qualitative
Which parts of the code do developers
talk about the most?
13

Case Study: Eclipse 2.0 - 3.2

A
Change-Log Analysis
B
Fuzzy Code Search

14


A
Change-Log Analysis
16,722 23,079
B
Fuzzy Code Search

Discussions Files

14


A
Change-Log Analysis
16,722 23,079
B
Fuzzy Code Search
3,865 13,581
Discussions Files Discussions Files

14


A
Change-Log Analysis
16,722 23,079
B
Fuzzy Code Search
3,865 13,581
Discussions Files Discussions Files

Q: Is A a Superset of B ?
14

Bug Reports (211,843)

15


Bug Reports containing
Code (33,301)

16


Code (33,301)

A
Clone Detection
Based Approach
(2,980)

17


Code (33,301)

A B
Clone Detection Change Log Analysis
Based Approach Based Approach
(2,980) (16,722)

18


Code (33,301)

A B
Based Approach 507 Based Approach
(2,980) (16,722)

19


Code (33,301)

A B
Based Approach 507 Based Approach

sa re
(2,980) (16,722)

p: Link
ov e rla nt!
al l iffere
ry sm ually d
Ve ce pt
c on
19

of c ode
y re-use ts.
e: h eav sion tes
I n Eclips reg res
o cr eate
ppe ts t
sni

20

Fuzzy Code Search

21

Fuzzy Code Search
Finds conceptually different links:

21

Fuzzy Code Search
What Developers really talk about

21

Fuzzy Code Search
Not what they end up changing

21

Fuzzy Code Search
Not what they end up changing
Not what their code references

21

Using Fuzzy Code Search to Link Code Fragments in Discussions to Source Code

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to Using Fuzzy Code Search to Link Code Fragments in Discussions to Source Code (20)

More from Nicolas Bettenburg (15)

Recently uploaded (20)

Using Fuzzy Code Search to Link Code Fragments in Discussions to Source Code