automatically generated patches as debugging aids

65
Automatically Generated Patches as Debugging Aids: A Human Study Yida Tao , Jindae Kim, Sunghun Kim Dept. of CSE, The Hong Kong University of Science and Technology Chang Xu State Key Lab for Novel Software Technology, Nanjing University

Upload: yida-tao

Post on 29-Jul-2015

198 views

Category:

Presentations & Public Speaking


0 download

TRANSCRIPT

Page 1: Automatically generated patches as debugging aids

Automatically Generated Patches as Debugging Aids: A

Human Study

Yida Tao, Jindae Kim, Sunghun Kim

Dept. of CSE, The Hong Kong University of Science and Technology

Chang Xu

State Key Lab for Novel Software Technology, Nanjing University

Page 2: Automatically generated patches as debugging aids

• Promising research progress• ClearView1: Prevent all 10 Firefox

exploits• GenProg2: Fix 55/105 real bugs

[1] Automatically Patching Errors in Deployed Software. Perkins et al. SOSP’09[2] A systematic study of automated program repair: fixing 55 out of 105 bugs for $8 each. Le Goues et al. ICSE’122

Automatic Program Repair

Page 3: Automatically generated patches as debugging aids

3

Automatic Program Repair

Page 4: Automatically generated patches as debugging aids

4

- Slashdot discussion: http://science.slashdot.org/story/09/10/29/2248246/Fixing-Bugs-But-Bypassing-the-Source-Code

“It won't get your bug patched any quicker. You’ll just have shifted the coders' attention away from their own app's bugs, and onto the repair tool’s bugs.”

Automatic Program Repair

Page 5: Automatically generated patches as debugging aids

#what-could-possibly-go-wrong• Blackbox repair• Increasing maintenance

cost• Vulnerable to attack

- Slashdot discussion: http://science.slashdot.org/story/09/10/29/2248246/Fixing-Bugs-But-Bypassing-the-Source-Code- A human study of patch maintainability. ISSTA’12- Automatic patch generation learned from human-written patches. ICSE’13

5

Page 6: Automatically generated patches as debugging aids

- Slashdot discussion: http://science.slashdot.org/story/09/10/29/2248246/Fixing-Bugs-But-Bypassing-the-Source-Code- A human study of patch maintainability. ISSTA’12- Automatic patch generation learned from human-written patches. ICSE’13

#program-out-of-control

6

#what-could-possibly-go-wrong• Blackbox repair• Increasing maintenance

cost• Vulnerable to attack

Page 7: Automatically generated patches as debugging aids

7

Use automatically generated patches as debugging aids

Page 8: Automatically generated patches as debugging aids

8

Use automatically generated patches as debugging aids

Our Human Study• Investigate the usefulness of

generated patches as debugging aids

• Discuss the impact of patch quality on debugging performance

• Explore practitioners’ feedback on adopting automatic program repair

Page 9: Automatically generated patches as debugging aids

9

Methodology

Page 10: Automatically generated patches as debugging aids

10

Bugs

Participants

Debugging aid

Debugis given to

Page 11: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 11

Page 12: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 12

Low-quality generated patch

Page 13: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 13

Low-quality generated patch

High-quality generated patch

Page 14: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 14

Low-quality generated patch

High-quality generated patch

Buggy method location

Page 15: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 15

Grad: 44

Engr: 28

MTurk: 23

95 Participants

CS graduate students

Industrial software engineers

Amazon Mechanical Turk workers

Page 16: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 16

Page 17: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 17

44 Graduate students• Between-group

design

14 students

15 students

15 students

Page 18: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 18

44 Graduate students• Between-group

design

Low-quality generated patch

High-quality generated patch

Buggy method location

14 students

15 students

15 students

Page 19: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 19

44 Graduate students• Between-group

design• Onsite setting• Eclipse IDE• Supervised session

Low-quality generated patch

High-quality generated patch

Buggy method location

14 students

15 students

15 students

Page 20: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 20

Low-quality generated patch

High-quality generated patch

Buggy method location

Remote participants(28 Engr + 23 MTurk)• Within-group design

Page 21: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 21

Remote participants(28 Engr + 23 MTurk)• Within-group design• Online debugging system

Low-quality generated patch

High-quality generated patch

Buggy method location

Page 22: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 22

Page 23: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 23

Bug Selection Criteria

• Real bugs• The bug has accepted patches written by developers• Proper number of bugs• The bug has generated patches with different quality

Page 24: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 24

Automatic patch generation learned from human-written patches. Kim et al. ICSE’13

Page 25: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 25

Automatic patch generation learned from human-written patches. Kim et al. ICSE’13

Auto-generated patch A Auto-generated patch B

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) if(sub!=null){ args[i+1] = sub.toString(); }}

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) args[parenCount+1] = new Integer(reImpl.leftContext.length);}

Page 26: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 26

Automatic patch generation learned from human-written patches. Kim et al. ICSE’13

Auto-generated patch A Auto-generated patch B

avg. ranking from 85 devs and students

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) if(sub!=null){ args[i+1] = sub.toString(); }}

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) args[parenCount+1] = new Integer(reImpl.leftContext.length);}

1.6

2.8

Page 27: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 27

Automatic patch generation learned from human-written patches. Kim et al. ICSE’13

Auto-generated patch A Auto-generated patch B

avg. ranking from 85 devs and students

High-Quality Patch Low-Quality patch

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) if(sub!=null){ args[i+1] = sub.toString(); }}

for (int i=0; i<parenCount; i++) SubString sub = (SubString)parens.get(i) args[parenCount+1] = new Integer(reImpl.leftContext.length);}

1.6

2.8

Page 28: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 28

Page 29: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 29

Participants submit 337 patches as their debugging outcome

Page 30: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 30

Participants submit 337 patches as their debugging outcome

Location109

LowQ112

HighQ116# submitted

patches w.r.t debugging aid

Page 31: Automatically generated patches as debugging aids

BugsParticipantsDebugging aid 31

Participants submit 337 patches as their debugging outcome

Location109

LowQ112

HighQ116# submitted

patches w.r.t debugging aid

Bug166

Bug274

Bug359

Bug476

Bug562

# submitted patches w.r.t bugs

Page 32: Automatically generated patches as debugging aids

32

Evaluation of debugging performance

Page 33: Automatically generated patches as debugging aids

Patch CorrectnessCorrectness

33

Page 34: Automatically generated patches as debugging aids

Patch Correctness

• Passing test casesCorrectness

34

Page 35: Automatically generated patches as debugging aids

Patch Correctness

• Passing test cases• Matching the semantics of original

accepted patches

Correctness

35

Page 36: Automatically generated patches as debugging aids

Patch Correctness

• Passing test cases• Matching the semantics of original

accepted patches• 3 evaluators

Correctness

36

Page 37: Automatically generated patches as debugging aids

Debugging Time

• Eclipse Plug-in•Website Timer

Correctness

Debugging time

37

Page 38: Automatically generated patches as debugging aids

Correctness

Debugging time

• Independent variables• Debugging aids• Bugs• Participant types• Programming

experience

38

Page 39: Automatically generated patches as debugging aids

Multiple Regression Analysis

Correctness

Debugging time

• Independent variables• Debugging aids• Bugs• Participant types• Programming

experience

correctness=α0+α 1 ∙ x1+α 2 ∙ x2+α 3 ∙ x 3+α 4 ∙ x4d ebugging time=β0+β1 ∙ x1+β2 ∙ x2+β3 ∙ x3+β 4 ∙ x4

39

Page 40: Automatically generated patches as debugging aids

Post-study Survey

• Helpfulness of debugging aids• Difficulty of bugs• Opinions on using generated patches as

debugging aids

Correctness

Debugging time

Survey feedback

40

Page 41: Automatically generated patches as debugging aids

41

Results

Page 42: Automatically generated patches as debugging aids

42

High-quality patches significantly improve debugging correctness

1

48%

33%

71%

Page 43: Automatically generated patches as debugging aids

43

High-quality patches significantly improve debugging correctness

1

48%

33%

71%

Lo-catio

n

LowQ HighQ

% of correct patches

48%

71%

Page 44: Automatically generated patches as debugging aids

44

Lo-catio

n

LowQ HighQ

% of correct patches

High-quality patches significantly improve debugging correctness

1

Positive Coefficient =

1.25p-value= 0.00 < 0.05

48%

71%

Page 45: Automatically generated patches as debugging aids

45

Lo-catio

n

LowQ HighQ

% of correct patches

Low-quality patches slightly undermine debugging correctness

2

48%

33%

71%

Page 46: Automatically generated patches as debugging aids

46

Lo-catio

n

LowQ HighQ

% of correct patches

Low-quality patches slightly undermine debugging correctness

2

Negative Coefficient = -0.55p-value= 0.09 48%

33%

71%

Page 47: Automatically generated patches as debugging aids

47

Lo-catio

n

LowQ HighQ

% of correct patches

Low-quality patches can undermine debugging correctness

2

Negative Coefficient = -0.55p-value= 0.09 48%

33%

71%

Page 48: Automatically generated patches as debugging aids

48

High-quality patches are more useful for difficult bugs3

Page 49: Automatically generated patches as debugging aids

49

High-quality patches are more useful for difficult bugs3

2

3

4

5

Bug Difficulty

Bug1Math-280

Bug2Rhino-114493

Bug3Rhino-192226

Bug4Rhino-217379

Bug5Rhino-76683

Page 50: Automatically generated patches as debugging aids

50

High-quality patches are more useful for difficult bugs3

Bug1 Bug2 Bug3 Bug4 Bug50%

10%

20%

30%

40%

50%

60%

70%

80%

90%

% of correct patches

Location LowQ HighQ

2

3

4

5

Bug Difficulty

Bug1Math-280

Bug2Rhino-114493

Bug3Rhino-192226

Bug4Rhino-217379

Bug5Rhino-76683

Page 51: Automatically generated patches as debugging aids

51

4The type of debugging aid does not affect debugging time

Page 52: Automatically generated patches as debugging aids

52

4The type of debugging aid does not affect debugging time

Location LowQ HighQ0

20

40

Debugging time (min)

Page 53: Automatically generated patches as debugging aids

53

5Other factors’ impact on debugging performance

Difficult bugs significantly slow down debuggingEngr and MTurk are more likely to debug correctly Novices tend to benefit more from HighQ patches

Page 54: Automatically generated patches as debugging aids

Helpfulness of de-bugging aidsVery

helpful

Helpful

Medium

Slightly Helpful

Not Helpful

54

Participants consider high-quality generated patches much more helpful than low-quality patches

Low-quality generated patch

High-quality generated patch

Mann-Whitney U testp-value =

0.001

6

Page 55: Automatically generated patches as debugging aids

55

Feedback

Page 56: Automatically generated patches as debugging aids

56

Page 57: Automatically generated patches as debugging aids

57

Quick starting point• Point to the buggy

area• Brainstorm

“They would seem to be useful in helping find various ideas around fixing the issue, even if the patch isn’t always correct on its own.”

Page 58: Automatically generated patches as debugging aids

58

Quick starting point• Point to the buggy

area• Brainstorm

Confusing, incomplete, misleading• Wrong lead, especially for

novices• Require further human

perfection“They would seem to be useful in helping find various ideas around fixing the issue, even if the patch isn’t always correct on its own.”

Page 59: Automatically generated patches as debugging aids

59

“Generated patches would be good at recognizing obvious problems”

“…but may not recognize more involved defects.”

Page 60: Automatically generated patches as debugging aids

60

“Generated patches would be good at recognizing obvious problems”

“…but may not recognize more involved defects.”

“Generated patches simplify the problem”

“…but they may over-simplify it by not addressing the root cause.”

Page 61: Automatically generated patches as debugging aids

61

“I would use generated patches as debugging aids, as they provide extra diagnostic information”

Page 62: Automatically generated patches as debugging aids

62

“I would use generated patches as debugging aids, as they provide extra diagnostic information”

“…along with access to standard debugging tools.”

Page 63: Automatically generated patches as debugging aids

63

Threats to Validity

Page 64: Automatically generated patches as debugging aids

64

Threats to Validity

• Bugs and generated patches may not be representative• Quality measure of generated patches may not

generalize• May not generalize to domain experts• Possibility of blindly reusing generated patches• Remove patches that are submitted less than 1 minute

Page 65: Automatically generated patches as debugging aids

65

Takeaway

• Auto-generated patches can be useful as debugging aids• Participants fix bugs more correctly

with auto-generated patches

• Quality control is required• Participants’ debugging correctness is

compromised with low-quality generated patches

• Maximize the benefits• Difficult bugs• Novice developers