SOFTWARE VISUALIZATION Lauren Wilkinson Shahar Maoz Picasso Bhowmik

SOFTWARE VISUALIZATION Lauren Wilkinson Shahar Maoz Picasso Bhowmik 1

What is Software Visualization? n Software visualization categorization* ¨ Algorithm visualization n n ¨ Static (flow chart) Animation Program visualization n n Static code / data viz (e. g. UML, ERD) Code / data animation (execution) * Stasko et al, 1998 2

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 ¨ 3

Things to think about n Key challenges* ¨ Scale n Existing applications work well on toy programs ¨ Transition n n between levels Code, control flow, class diagram, package, application Design time, run time ¨ Automation n Layout * Stasko et al, 1998 4

Algorithm Animations n Algorithms in Action – A program for learning basic algorithms, developed by Linda Stern, Lee Naish, and Harald Sondergaard, at The University of Melbourne. 5

BUBBLE SORT n n 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/Bubble. Sort. html 6

Selection 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/Selection. Sort. html 7

Quicksort n 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/Quick. Sort. html 8

2 -3 -4 Tree n 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). n 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/Tree 234. html 9

Minimum Spanning Tree Algorithm n n 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. html 10

Flow Diagrams n The two most important modeling techniques used in analyzing and building information systems are : – ¨Data Flow Diagrams (DFDs) ¨Entity-Relationship Diagrams (ERDs) 11

Data Flow Diagrams (Dfds) n DFD Principles n Basic DFD Notations n ERD Principles n Basic ERD Notations 12

What processing is done? When? How? Where? By which component? What data is needed? By which component? for what? When? 13

Principles System can be decomposed into subsystems. n Subsystem represents a process or activity in which data is processed. n Each 'process' in a DFD has the characteristics of a system. n Process must have input and output. n Data Input – Data Flows – Data Output n 14

Example 15

Entity Relationship Diagrams A simple entity-relationship diagram 16

Principles There are three basic elements in ER models: n. Entities are the "things" about which we seek information. n. Attributes are the data we collect about the entities. n. Relationships provide the structure needed to draw information from multiple 17 entities.

The Unified Modeling Language (UML) n n n 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 18

Issues with UML n Class Diagrams Aesthetics and Usability: empirical study* ¨ Goal n ¨ Procedure n n n ¨ Identify the most important aesthetics for the automatic layout of UML class diagrams from a human comprehension point of view 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 n n “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 19

20

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/cs 263/paper. html 21

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. , 22 Goldberg, A. , and Lewis, T. , editors, Visual Object-Oriented Programming. Prentice-Hall, New York, 1994.

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. , 23 Goldberg, A. , and Lewis, T. , editors, Visual Object-Oriented Programming. Prentice-Hall, New York, 1994.

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. , 24 Goldberg, A. , and Lewis, T. , editors, Visual Object-Oriented Programming. Prentice-Hall, New York, 1994.

Prograph: A Completely Iconic Programming Language http: //www. mactech. com/articles/mactech/Vol. 10/10. 11/Prograph. CPXTutorial/ 25

Loops in Prograph http: //www. mactech. com/articles/mactech/Vol. 10/10. 11/Prograph. CPXTutorial/ 26

Cube: Function for the Factorial of a Number 27 Najork & Kaplan 1991 Najork, M. and Kaplan, S. The cube language. In Proc. 1991 IEEE Workshop Visual Languages , pp. 218 -224, Kobe, Japan, 1991.

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 28

Foam: A Java Swing Developer http: //www. computersinmotion. com 29

See. Soft: a look at the source code* n n n 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 2 D and a new application in 3 D * Eick et al, 1995 30

31

32

Building on the See. Soft Metaphor: Source Viewer 3 D (sv 3 D) • 3 D representation for visualizing large software systems • Extends the See. Soft metaphor • Applications include: • fault localization (debugging) • visualization of execution traces • source code browsing 33 Marcus et al. 2003 Marcus, A. , Feng, L. , and Maletic, J. 3 D Representations for Software Visualization. ACM Symposium on Software Visualization, San Diego.

Customized Program Visualizations in sv 3 D • Users can define mappings between software elements and visualization components • Data can be mapped to visual elements of color, position, height and depth 34 Marcus et al. 2003 Marcus, A. , Feng, L. , and Maletic, J. 3 D Representations for Software Visualization. ACM Symposium on Software Visualization, San Diego.

Viewing Complex Source Code with sv 3 D 2 D Overview of Source Code 3 D Overview of Source Code Color represents nesting level • Each cylinder is a line of source code • Color represents control structure type Color represents control structure • Height represents nesting level 35 Marcus et al. 2003 Marcus, A. , Feng, L. , and Maletic, J. 3 D Representations for Software Visualization. ACM Symposium on Software Visualization, San Diego.

Overcoming Occlusion in sv 3 D Elevation Transparency 36 Marcus et al. 2003 Marcus, A. , Feng, L. , and Maletic, J. 3 D Representations for Software Visualization. ACM Symposium on Software Visualization, San Diego.

Tarantula: Localizing Program Faults Food for Thought: Could sv 3 D improve this visualization? http: //www. cc. gatech. edu/aristotle/Tools/tarantula/index. html 37

Design Exercise Design a better visualization for the UML diagram handed out and discussed by Shahar. n Remember: Maintain the same logic n Encouraged: Use 3 D, Animation, Colors n 38