SOFTWARE VISUALIZATION

1
SOFTWARE VISUALIZATION

• Lauren Wilkinson
• Shahar Maoz
• Picasso Bhowmik

2
What is Software Visualization?

• Software visualization categorization
• Algorithm visualization
• Static (flow chart)
• Animation
• Program visualization
• Static code / data viz (e.g. UML, ERD)
• Code / data animation (execution)

Stasko et al, 1998
3
What to visualize?

• Source code and configuration data and text,
  metadata, file and function size, access history,
  bug locations and fixes
• Code and data structure data structures, data
  flow, function calls
• Execution data and algorithms animation, message
  passing, control flow, memory and resource
  utilization
• Code and human interaction development process

4
Things to think about

• Key challenges
• Scale
• Existing applications work well on toy programs
• Transition between levels
• Code, control flow, class diagram, package,
  application
• Design time, run time
• Automation
• Layout

Stasko et al, 1998
5
Algorithm Animations

• Algorithms in Action A program for learning
  basic algorithms, developed by Linda Stern, Lee
  Naish, and Harald Sondergaard, at The University
  of Melbourne.

6
BUBBLE SORT

• Simplest sorting method.
• For a list of size n the algorithm passes
  through the list n-1 times. At each pass, every
  two adjacent elements that are not in correct
  order are swapped.

http//www.csse.monash.edu.au/dwa/MELB/BubbleSort
.html
7
Selection Sort

• Sorts by repeatedly choosing the largest item
  among the unsorted items, and exchanging it with
  the item in its correct position.

http//www.csse.monash.edu.au/dwa/MELB/SelectionS
ort.html
8
Quicksort

• Recursively partitions an array around a
  partition element (Divide Conquer)
• One partition contains elements less than or
  equal to the partition element,
• Other partition contains elements greater than or
  equal to the partition element.

http//www.cs.mu.oz.au/aia/QuickSort.html
9
2-3-4 Tree

• In the 2-3-4 tree, nodes can contain one key
  (plus links to 2 children, so called 2-nodes),
  two keys (3-nodes), or three keys (4-nodes).
• New items are always inserted into already
  existing leaf nodes, converting 2-nodes to
  3-nodes, and 3-nodes to 4-nodes.

http//www.cs.mu.oz.au/aia/Tree234.html
10
Minimum Spanning Tree Algorithm

• Work by choosing a minimum cost edge at each
  step.
• In Prim's algorithm, we start with an arbitrarily
  chosen vertex as the root of a tree T, and at
  each step we add to T the edge e VW of minimum
  cost, where V is already in T and W is not in T.

http//www.csse.monash.edu.au/dwa/MELB/Spanning.h
tml
11
Flow Diagrams

• The two most important modeling techniques used
  in analyzing and building information systems are
  
• Data Flow Diagrams (DFDs)
• Entity-Relationship Diagrams (ERDs)

12
Data Flow Diagrams (Dfds)

• DFD Principles
• Basic DFD Notations
• ERD Principles
• Basic ERD Notations

13

• What processing is done?  When? How? Where? By
  which component?
• What data is needed?  By which component? for
  what? When?

14
Principles

• System can be decomposed into subsystems.
• Subsystem represents a process or activity in
  which data is processed.
• Each 'process' in a DFD has the characteristics
  of a system.
• Process must have input and output.
• Data Input Data Flows Data Output

15
Example
16
Entity Relationship Diagrams
A simple entity-relationship diagram
17
Principles

• There are three basic elements in ER models
• Entities are the "things" about which we seek
  information.
• Attributes are the data we collect about the
  entities.
• Relationships provide the structure needed to
  draw information from multiple entities.

18
The Unified Modeling Language (UML)

• A modeling language for specifying, visualizing,
  constructing, and documenting systems
• Based on the Object Oriented Paradigm
• Accepted as industry standard (1997)
• Nine types of diagrams class, object, use case,
  sequence, collaboration, statechart, activity,
  component, and deployment.
• Examples from Ideogramic

19
Issues with UML

• Class Diagrams Aesthetics and Usability
  empirical study
• Goal
• Identify the most important aesthetics for the
  automatic layout of UML class diagrams from a
  human comprehension point of view
• Procedure
• CS students were given short UML class with
  examples, then read a textual spec and asked to
  identify correct / incorrect class diagrams
• Diagrams shown in random order, each with
  specific aesthetic metric value set to very high
  or very low (bends, orthogonality, edge
  variation, node distribution, direction of flow)
• Correctness and time-to-response measured
• Conclusions
• Tempting to say that none of the aesthetics
  really matters
• Domain specific algorithms are required
• Semantic grouping of related objects (e.g.
  position subclasses in an inheritance hierarchy
  close to each other)
• A nice layout is unlikely to be sufficient for
  intuitive use

Purchase et al, 2001
20
(No Transcript)
21
Visual Programming Languages

• A purely visual language relies completely on
  visualization
• No textual representation at all
• The programmer manipulates icons or other
  graphical representations to create a program,
  which is then executed and debugged in the same
  visual environment
• Examples VIPR, Prograph, Cube

Marat Boshernitsan and Michael Downes. Visual
Programming Languages A Survey. CS Division, UC
Berkeley. http//www.cs.berkeley.edu/maratb/cs263
/paper/paper.html
22
VIPR Visualization of Program Execution
Citrin et al. 1994 Citrin, W., Doherty, M., and
Zorn, B. Design of a completely visual
object-oriented programming language. In Burnett,
M., Goldberg, A., and Lewis, T., editors, Visual
Object-Oriented Programming. Prentice-Hall, New
York, 1994.
23
VIPR Control Statements and while loops
Citrin et al. 1994 Citrin, W., Doherty, M., and
Zorn, B. Design of a completely visual
object-oriented programming language. In Burnett,
M., Goldberg, A., and Lewis, T., editors, Visual
Object-Oriented Programming. Prentice-Hall, New
York, 1994.
24
VIPR Function Calls
Citrin et al. 1994 Citrin, W., Doherty, M., and
Zorn, B. Design of a completely visual
object-oriented programming language. In Burnett,
M., Goldberg, A., and Lewis, T., editors, Visual
Object-Oriented Programming. Prentice-Hall, New
York, 1994.
25
Prograph A Completely Iconic Programming Language
http//www.mactech.com/articles/mactech/Vol.10/10.
11/PrographCPXTutorial/
26
Loops in Prograph
http//www.mactech.com/articles/mactech/Vol.10/10.
11/PrographCPXTutorial/
27
Cube Function for the Factorial of a Number
Najork Kaplan 1991 Najork, M. and Kaplan, S.
The cube language. In Proc. 1991 IEEE Workshop
Visual Languages, pp. 218-224, Kobe, Japan, 1991.
28
Graphical Development Tools

• Related to visual programming, but the language
  itself is not visual
• Textual languages with a graphical interface
• Great for layout/GUI development
• Examples Foam, Visual Basic, Dreamweaver

29
Foam A Java Swing Developer
http//www.computersinmotion.com
30
SeeSoft a look at the source code

• Visualizes text files by mapping each line into a
  thin row, colored according to a statistic of
  interest. Any text and any statistics about the
  text may be used.
• Interesting case is source code with statistics
  such as the age, programmer, or functionality of
  each line. These statistics are derived from a
  variety of sources, such as version control
  systems, static analysis, and profiling.
• Examples in 2D and a new application in 3D

Eick et al, 1995
31
(No Transcript)
32
(No Transcript)
33
Building on the SeeSoft Metaphor Source Viewer
3D (sv3D)

• 3D representation for visualizing large software
  systems
• Extends the SeeSoft metaphor
• Applications include
• fault localization (debugging)
• visualization of execution traces
• source code browsing

Marcus et al. 2003 Marcus, A., Feng, L., and
Maletic, J. 3D Representations for Software
Visualization. ACM Symposium on Software
Visualization, San Diego.
34
Customized Program Visualizations in sv3D

• Users can define mappings between software
  elements and visualization components
• Data can be mapped to visual elements of color,
  position, height and depth

Marcus et al. 2003 Marcus, A., Feng, L., and
Maletic, J. 3D Representations for Software
Visualization. ACM Symposium on Software
Visualization, San Diego.
35
Viewing Complex Source Code with sv3D
2D Overview of Source Code
3D Overview of Source Code
Color represents nesting level

• Each cylinder is a line of source code
• Color represents control structure type
• Height represents nesting level

Color represents control structure
Marcus et al. 2003 Marcus, A., Feng, L., and
Maletic, J. 3D Representations for Software
Visualization. ACM Symposium on Software
Visualization, San Diego.
36
Overcoming Occlusion in sv3D
Elevation
Transparency
Marcus et al. 2003 Marcus, A., Feng, L., and
Maletic, J. 3D Representations for Software
Visualization. ACM Symposium on Software
Visualization, San Diego.
37
Tarantula Localizing Program Faults
Food for Thought Could sv3D improve this
visualization?
http//www.cc.gatech.edu/aristotle/Tools/tarantula
/index.html
38
Design Exercise

• Design a better visualization for the UML diagram
  handed out and discussed by Shahar.
• Remember Maintain the same logic
• Encouraged Use 3D, Animation, Colors