Presentation is loading. Please wait.

Presentation is loading. Please wait.

Software Visualization Using Hierarchical Edge Bundles and Massive Sequence Views Danny Holten and Jarke J. van Wijk Eindhoven University of Technology.

Similar presentations


Presentation on theme: "Software Visualization Using Hierarchical Edge Bundles and Massive Sequence Views Danny Holten and Jarke J. van Wijk Eindhoven University of Technology."— Presentation transcript:

1 Software Visualization Using Hierarchical Edge Bundles and Massive Sequence Views Danny Holten and Jarke J. van Wijk Eindhoven University of Technology Technische Universiteit Eindhoven May 13, 2009 Eindhoven

2 Motivation  Program Comprehension  Purpose: maintenance, change requests  Challenge: static and/or dynamic analysis  Challenge: scalable visualization  Dynamic Analysis  Accurate picture of what is actually happening  Late binding  Object identities  Execution Traces  Execution of a (feature) scenario

3 Problem Statement  Visualization – Properties  Many structural elements  Many interrelationships  Typically: UML Sequence Diagrams (SD’s)  Visualization – Purpose  Exploration, feature location, and understanding  Straightforward Approaches  Do not suffice  SD’s: too much scrolling, too much wasted space

4 Novel Visualization Approach  Combination of Two Views  Bundle view based on previous work  Generalization of SD’s  Hierarchical Edge Bundles (HEB’s)  Visualize relations as bundles on top of hierarchy  Massive Sequence View (MSV)  Show order of calls as SD’s with improved scaling

5 HEBs – Data to Visualize  Hierarchy  Tree structure extracted from program  For Java: Package – Class – Method  Analogous for other languages  Adjacency Relations  Class-to-class calls  Method-to-method, package-to-package…  Analogous for other languages

6 HEB’s – Data to Visualize

7 HEBs – Main Idea

8 Sequence Diagrams  Interaction between system components  Calls time-ordered along vertical axis  Do not scale well due to graphical depiction

9 Massive Sequence View  Interaction between system components  Calls time-ordered along vertical axis  Scales well due to improved graphical depiction and rendering (IBAA)  Zoomable (Level-of-Detail)

10 Importance-Based Anti-Aliasing

11 Integration of Multiple Views: ExtraVis

12 Final Remarks on ExtraVis  Limitations  Performance diminishes above 500.000 calls  Large traces demand a lot of computational resources  Threads not (yet) supported  Project Location  http://swerl.tudelft.nl/extravis and  http://www.win.tue.nl/~dholten (more info on HEBs as well)  Tool download  High-resolution screenshots

13 Visual Comparison of Hierarchically Organized Data Using Hierarchical Edge Bundles

14 Why Compare Trees?  Understand relationship between trees  High-level overview of correspondence / differences  Splits, joins, relocations of substructures  Elements unique to each tree  Various areas: evolutionary biology, HR & personnel, software development  Software development  Quickly gain insight in software architecture  Overview of differences and similarities between versions  Identify code reorganization 14

15 Visualization Problem  Show both trees… 15

16 Visualization Problem  Show both trees… 16

17 Visualization Problem  …and correspondence between trees 17

18 Visualization Problem  Improvement: icicle plots with straightened leafs 18

19 Visualization Problem  Basic visualization using straight lines 19

20 Crossing Reduction 20

21 Crossing Reduction 21

22 Crossing Reduction 22

23 Crossing Reduction 23

24 Crossing Reduction 24

25 Crossing Reduction 25

26 Crossing Reduction 26

27 Crossing Reduction 27

28 Crossing Reduction 28

29 Crossing Reduction 29

30 Hierarchical Edge Bundles  Introduce HEBs to replace straight lines  “Hierarchical Edge Bundles: Visualization of Adjacency Relations in Hierarchical Data”, IEEE InfoVis 2006 30

31 Hierarchical Edge Bundles  Bundling edges using the available hierarchy  “Hierarchical Edge Bundles: Visualization of Adjacency Relations in Hierarchical Data”, IEEE InfoVis 2006 31

32 Hierarchical Edge Bundles  Tree structures are “mirrored” to the center to provide HEB guides 32

33 Hierarchical Edge Bundles  Different bundling strengths can be chosen 33

34 Hierarchical Edge Bundles  Bundling controllable per hierarchy level 34

35 Program Demo

36 Final Remarks on Tree Comparison  Advantageous features  High-level overview of correspondence  Hierarchies are sorted to facilitate comparison  Bundling controllable per hierarchy level  Direct interaction with correspondence using “crossing”  Disadvantages  Bundling can obscure individual relations  Availability/generation of matching list necessary  Non-atomicity of matching elements (granularity problem) 36

37 Questions?


Download ppt "Software Visualization Using Hierarchical Edge Bundles and Massive Sequence Views Danny Holten and Jarke J. van Wijk Eindhoven University of Technology."

Similar presentations


Ads by Google