Using Dynamic Analysis to Create
Trace-Focused User Interfaces for IDEs
University of Victoria
Victoria, British Columbia
University of Victoria
Victoria, British Columbia
This research demonstration presents the tool, Dynamic In-
teractive Views for Reverse Engineering (Diver). Diver sup-
ports software understanding through a trace focused user
interface. The trace focused user interface is a method of
re-organizing the user interface of integrated development
environments so that developers can focus their attention on
artifacts related to the run-time behaviour of the software
that they are investigating.
from research in software visualization, dynamic analysis,
software reconnaissance, and task focused user interfaces.
The tool combines concepts
Categories and Subject Descriptors
D.2.6 [Software Engineering]: Programming Environments
Reverse Engineering, Integrated Development Environments,
Software Reconnaissance, Visualization
The complexity of current software systems makes them
difficult to understand. Modern IDEs offer extensive sup-
port through advanced features such as searching for pro-
gram elements, free-form text queries, and visualizations of
type hierarchies. However, this support is limited by lan-
guage features such as polymorphism and dynamic bind-
ing. Dynamic analysis may be employed to overcome these
limitations. This method of analysis typically logs runtime
information of in a dynamic execution trace. Traces can be
analyzed to extract information about a running system and
visualized using UML sequence diagrams.
In our previous work , we conducted a study to de-
termine the requirements for supporting reverse engineer-
ing tasks using sequence diagrams within an integrated de-
velopment environment (IDE). We found that engineers re-
quire linked views of both the static and dynamic aspects of
their program to effectively understand their software. It is
particularly important for the views to be integrated with
Copyright is held by the author/owner(s).
FSE-18, November 7–11, 2010, Santa Fe, New Mexico, USA.
Interactive Tools for Reverse Engineering (Diver)1tool.
Diver provides a completely new perspective for a devel-
oper’s IDE, through what we refer to as a trace focused user
interface (UI). The user interactively creates a set of ex-
ecution traces that are displayed in a special view, called
the Program Traces View. Traces in this view can then be
selected by the user to both filter and highlight relevant in-
formation about the program execution. The user can now
interactively explore this information using standard IDE
views linked to our custom sequence diagram view and trace
search features. This trace focused UI approach is unique
because it is the first one to combine research in software vi-
sualization , dynamic analysis , software reconnaissance
 and task focused user interfaces .
Our findings led us to design the Dynamic
2.DIVER AND THE TRACE FOCUSED
Diver is implemented as a set of plug-ins for the Eclipse
integrated development environment.2It offers a perspec-
tive that gives the user several views which help him or her
interact with source code and dynamic traces. All of the
views are linked to provide a rich interaction experience for
the user. Each of the views are designed or modified to
allow users to gain insight about the runtime behaviour of
their software through the use of dynamic execution traces.
Execution traces are created by the user through the stan-
dard Eclipse launch facilities and through specialised actions
contributed to the Eclipse Debug View (Fig. 1-A).
Central to the Diver perspective is the Program Traces
View (Fig. 1-B) which shows execution traces created by
the user.The traces are organized by launch name and
the date and time of the trace. Each thread of execution
contained in the trace is also accessible from this view.
The user can use the Program Traces View to open a vi-
sualization of a thread in Diver’s custom Sequence Diagram
View (Fig. 1-C). Diver sequence diagrams are unique in
that they visualize runtime as well as source code informa-
tion. This is achieved through an algorithm which matches
method calls in the trace to the blocks of source code. These
matchings are used to display combined fragments that con-
tain method calls and to compact iterations of loops. Our
empirical study indicates that compacting loop iterations
can reduce the size of the sequence diagrams for industrial
software by up to 80% . The sequence diagram is also
Figure 1: The Views of Diver
fully interactive. Lifelines, combined fragments, and activa-
tion boxes can all be expanded or collapsed to increase or
decrease the amount of information that is viewed. Every
element in the view is selectable, and the view supports key-
board navigation. The view can also be navigated using a
linked outline view (Fig. 1-D).
The Program Traces view can also be used to “activate”
a trace. Activating a trace does several things to change
the Eclipse user interface. First, it provides a code coverage
visualization. The Eclipse Source Code Editor (Fig. 1-E) is
marked with annotations that indicate the lines of code on
which methods were called during the trace.
Activating a trace also initiates software reconnaissance.
Software reconnaissance is a method for locating features in
source code by comparing information logged in multiple ex-
ecution traces . Diver filters The Eclipse Package Explorer
(Fig. 1-F) so that only program elements that are referenced
in the active trace are visible. The user is also able to use
the Program Traces View to “hide” traces. This is done by
activating a trace and selecting one or more traces to hide
in the view. Hiding traces adjusts the Diver filter so that
only source code elements in the active trace, but not the
hidden traces, are visible. This enables users to locate fea-
tures in source code by focusing on program elements that
are unique to those features.
The Eclipse Package Explorer is also linked to the Diver
Sequence Diagram View. When a trace is active, users can
right-click on a class or method to reveal, in the Sequence
Diagram View, the first time it was referenced in a selected
thread in the trace. Other references to the selected method
or class can be revealed using a timeline widget which is
displayed at the very bottom of the sequence diagram. The
timeline is marked by vertical bars representing all of the
calls to the selected element in the Package Explorer.
Finally, every element in a trace may be annotated using
the standard Eclipse Properties View (Fig. 1-G). Annota-
tions can later be searched using the Diver Trace Search.
Wild-card searches for classes and methods are also sup-
Diver is implemented in Java and C++ and consists of
more than 35K classes. It uses the Java Virtual Machine
Tooling Interface3to generate traces of Java applications or
Eclipse Rich Client Platform (RCP) applications. Traces
contain data about method and constructor calls and re-
turns. Users can control which method and constructor calls
are stored by using a button supplied in the Eclipse Debug
View. The data is stored in binary files which are later an-
alyzed and indexed into a database. Information indexed
into the database can be controlled through customisable
“trace analysis filters”. By default, Diver only indexes in-
formation related to source code in the project that defines
the Java application, or the plug-ins that define the Eclipse
RCP application. The process is scalable in our experience.
We have used Diver to analyze large software applications
such as the Eclipse IDE, the Jetty HTTP Server4and the
4. CONCLUSIONS AND FUTURE WORK
Diver combines research on trace visualization, software
reconnaissance, and task focused user interfaces to create
a unique user interface for program exploration. We have
performed a user study in which participants were asked to
use Diver to locate and analyze features of Diver itself. The
results of this study are promising and have been submitted
for publication. We are also currently logging information
from current users of Diver to investigate how they use the
tool. Diver is available as open source from our web site:
 C. Bennett, D. Myers, M. Storey, D. German,
D. Ouellet, M. Salois, and P. Charland. A survey and
evaluation of tool features for understanding
reverse-engineered sequence diagrams. Journal of
Software Maintenance and Evolution: Research and
Practice, 20(4):291–315, 2008.
 B. Cornelissen, A. Zaidman, A. van Deursen,
L. Moonen, and R. Koschke. A systematic survey of
program comprehension through dynamic analysis.
IEEE Trans. on Software Engineering, 35(5):684 –702,
 M. Kersten and G. C. Murphy. Mylar: a degree-of-
interest model for ides. In Proc. of the 4th Int’l. Conf.
on Aspect-Oriented Software Development, pages
159–168, New York, USA, 2005. ACM.
 D. Myers, M.-A. Storey, and M. Salois. Utilizing debug
information to compact loops in large execution traces.
In Proc. of the 14th European Conference on Software
Maintenance and Re-engineering, pages 41–50. IEEE,
 N. Wilde and M. Scully. Software reconnaissance:
mapping program features to code. Journal of Software
Maintenance: Research and Practice, 7(1):49–62, 1995.