Web Application Fault Classification An Exploratory Study Yuepu - - PowerPoint PPT Presentation

Web Application Fault Classification — An Exploratory Study

Yuepu Guo, Sreedevi Sampath University of Maryland Baltimore County

presentation by Daniel Kellenberger 01.06.2010

2

Why Do We Need (Another) Fault Classification?

• Fault-based software testing
• Other classifications are…

— Not based on empirical data — Not designed for web applications

• ODC was used as guideline for the paper

3

Orthogonal Defect Classification (ODC)

• Opener section

— Activity — Trigger — Impact

• Closer section

— Target — Defect type — Qualifier — Age — Source

4

Web Applications

• Large user group
• Heterogeneous…

— Components — Technologies — Programming languages — etc…

• Multiple entry points

5

Build a Taxonomy

• Select a dimension
• Specify baseline
• Add/adjust categories

6

Dimension

• Traditional ODC categories
• Web specific categories

— server platform — web browser — etc…

Physical location

• f a fault

7

Baseline

• Data store faults
• Logic faults
• Form faults
• Appearance faults
• Link faults

8

Criteria of the selected Applications

• From sourceforge.net
• Web-based Java application
• More than 10K LOC
• Activity index greater than 70%

9

Roller Weblogger

10

qaManager

11

Selected Applications

Business Communications Domain 161 104 Total bugs 12 3 Open bugs 49030 32848 LOC 99.9% 79.1% Activity Index 9 5 Developers qaManager Roller Weblogger

12

Fault Analysis Procedure

• Analyze the bugs

— Look at the bug report — If necessary look into the code

• Try to assign to a existing category

— Create new category — Split category into sub categories

13

Results

100% 61.5% 3.7% 2.5% 5.6% 14.9% 11.8% 100% 59.6% 2.9% 1.0% 12.5% 12.5% 11.5% 161 104 Total Faults 99 62 Logic 6 3 Compatibility 4 1 Form 9 13 Link 24 13 Appearance 19 12 Data Store qaManager Roller Weblogger Number of Faults Fault Category

14

Results cont.

100% 86.0% 3.0% 2.0% 0.0% 2.0% 3.0% 4.0% 100% 56.5% 8.1% 6.4% 9.7% 8.1% 4.8% 6.4% 4 4 Browser interaction 3 3 Paging 2 5 Session 6 Server-side parsing 2 4 Encoding/decoding 3 5 Locale 85 35 Other 99 62 Total Logic Faults qaManager Roller Weblogger Number of Faults Fault Category

15

Discussion

• They say

— Hope that other researchers will use their fault classification — Plan to do larger study (other than java)

• My opinion

— Just a proposition — Only 2 applications (and only java) — Only little content — Security aspect (no category for that)

Questions?