Issue trackers have barely changed.

Apps have.

Tap
Long touch
Swipe
Type
Describe the bug…

Touch-driven event flows, squeezed into one text box.

0%

of users are less likely to use an app again
when it doesn’t perform as expected.

MASTER’S THESIS · 2015

Enhancing Bug Reports for Mobile Apps

Kevin Moran
College of William & Mary
Department of Computer Science
FUSION, in full: the approach, two studies, every result.
CONTENTS
1Introduction
2Related Works
3The FUSION Approach
4Design of Empirical Studies
5Empirical Study Results
6Limitations & Threats
7Conclusion
CHAPTER 3

Analyze, then generate.

Program analysis learns the app once; reporters are guided through every step.

ANALYSIS
REPORT GENERATION
1

Primer

static analysis

2

Engine

dynamic DFS ripping

3

Database

components · screens

4

Auto-completion

suggests the next step

5

Report entry

testers write steps

6

Reports

developers reproduce

CHAPTER 4

Two studies. Real bugs.

15 bugs from 14 open-source Android apps, against Google Code Issue Tracker.

STUDY 1 · REPORTING

Write the reports

0reporters, half inexperienced
0reports · 60 per system
→
STUDY 2 · REPRODUCING

Reproduce the bugs

0new participants
0reports incl. 15 originals
CHAPTER 5

Detail takes longer.

Average time to write a bug report.

0:00
with FUSION
0:00
with Google Code
CHAPTER 5

Then it pays off.

Unreproducible reports, by reporter experience.

EXPERIENCED
INEXPERIENCED
FUSION
0
0
Google Code
0
0

Enhancing Bug Reports for Mobile Apps

Kevin Moran
MS Thesis 2015 · College of William & Mary · Department of Computer Science