27
On the Central Role of Mailing Lists: An Exploratory Study Emad Shiahb Nicolas Bettenburg Bram Adams Ahmed E. Hassan

Kcsd2009 emad

  • Upload
    sailqu

  • View
    91

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Kcsd2009 emad

On the Central Role of Mailing Lists: An Exploratory Study

Emad ShiahbNicolas Bettenburg

Bram AdamsAhmed E. Hassan

Page 2: Kcsd2009 emad

The Central Role of Mailing Lists

Open Source developers are distributed ….

... they communicate through mailing lists

… this makes mailing lists a rich data source that we can leverage

Page 3: Kcsd2009 emad

The Central Role of Mailing Lists

Page 4: Kcsd2009 emad

What we Studied…

DevelopersExternalFactors

SourceCode

Q4. Competition?Q1. Communicationstyle?

Q2. Stability?

Q3. Code changes vs.Mailing list activity?

Page 5: Kcsd2009 emad

Mailing Lists Studied

Case study using 22 different GNOME projects

Page 6: Kcsd2009 emad

Participant Message Contribution

Question 1

Page 7: Kcsd2009 emad

40%

60%

Participant Message Contribution

Divided participants into 2 groups:

Dominant group(Top 10% contributors)

Casual group(The rest of contributors)

Page 8: Kcsd2009 emad

2%

82%

16%

Distribution of Participation in Threads

Dominant only

Dominant and Casual

Casual only

Page 9: Kcsd2009 emad

Participant Message Contribution

Question 1

Answer 1

Page 10: Kcsd2009 emad

Stability of Participants

Question 2

Page 11: Kcsd2009 emad

Stability of Dominant Group is Important!

2%

82%

16%Dominant Only

Dominant and Casual

Casual Only

Dominant group are involved in 98% of threads

Page 12: Kcsd2009 emad

Stability of Participants

Cosine Distance (CD) to measure participant stability over time

CD = 1 0 < CD < 1 CD = 0

Page 13: Kcsd2009 emad

Stability of Participants

Cosine Distance

Casual

Time

DominantAvg. = 0.69

Avg. = 0.15

0

Page 14: Kcsd2009 emad

Stability of Participants

Question 2

Answer 2

Page 15: Kcsd2009 emad

Mailing List and Source Code Activity

Question 3

Page 16: Kcsd2009 emad

Mailing List and Source Code Activity

Code Activity (CA) = Lines added + Lines Deleted + Lines modified

Mailing List Activity = No. of messages posted

Page 17: Kcsd2009 emad

Mailing List and Source Code Activity

76.0

Page 18: Kcsd2009 emad

Type of Source Code Change and Mailing List Activity

Divide the code changes into the 3 categories:1. Additions2. Removes3. Modifications

… and measure correlation with Mailing List activity ….

Page 19: Kcsd2009 emad

Type of Source Code Change and Mailing List Activity

Associate words to type of Code changes

“new features”“feature request”

Additions

“bug”“patch”

“testing”“maintain”

Modifications

Page 20: Kcsd2009 emad

Type of Source Code Change and Mailing List Activity

Additions terms mentioned 57% time more in Evolution compared to Nautilus

Modifications terms were mentioned in 75% more messages in Nautilus compared to Evolution

Consistent

Page 21: Kcsd2009 emad

Mailing List and Source Code Activity

Question 3

Answer 3

Page 22: Kcsd2009 emad

Effect of External Factors on Mailing List Activity

Question 4

Page 23: Kcsd2009 emad

Effect of External Factors on Mailing List Activity

Page 24: Kcsd2009 emad

Effect of External Factors on Mailing List Activity

Quantitative study: measure the occurrence of competing mailing clients to Evolution

Page 25: Kcsd2009 emad

Effect of External Factors on Mailing List Activity

Qualitative study: study mailing lists content

Page 26: Kcsd2009 emad

Effect of External Factors on Mailing List Activity

Question 4

Answer 4

Page 27: Kcsd2009 emad

The Central Role of Mailing Lists