a scientific analysis of opensuse collaboration patterns in obs and bugzilla - the story of merge...

Post on 04-Aug-2015

104 Views

Category:

Data & Analytics

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

A scientific analysis of openSUSE collaborationpatterns in OBS and Bugzilla

The story of merge requests and bug reports

Athanasios-Ilias Rousinopoulos<zoumpis@opensuse.org>

Alpen-Adria University

September 9, 2014

Introduction

Goals

• Mine openSUSE Factory data• Analyze the data• Correlate bugs with accepted requests• Publish the source code

3 of 30

Data

• Period of study: 3 years• Requests submitted to openSUSE Factory repository (via OBS)• Bugs for openSUSE Factory distribution in Bugzilla

4 of 30

Bugzilla

Mining process

• Bicho• 8 products• 6 projects [KDE,GNOME,Apache,BaseSystem,Kernel,Xfce]

6 of 30

Problem

7 of 30

OBS Factory

Mining process

• Download the OBS data in XML format• Develop an XML parser in Python• Parsed data are stored in MySQL database

9 of 30

Metrices

• Requests• Submitters (or Requesters)• Reviews• Reviewers• Packages• Accepters• Time deltas

10 of 30

Data Summary

Project Name Requests Reviews Reviewers PackagesGNOME:Factory 8355 10371 14 571KDE:Distro:Factory 6508 14144 16 340

11 of 30

GNOME:Factory

Requests per state

13 of 30

Submitters

14 of 30

Reviewers

15 of 30

Reviewed requests (accepted)

16 of 30

Accepters

17 of 30

Acceptance Time

18 of 30

KDE:Distro:Factory

Requests per state

20 of 30

Submitters

21 of 30

Reviewers

22 of 30

Reviewed requests (accepted)

23 of 30

Accepters

24 of 30

Acceptance Time

25 of 30

Source code

Get the source

• Alpha version• OBSParser

27 of 30

Conclusion

• Higher number bugs will lead us to better observations• The majority of requests are accepted in less than 10 days• People who submitt a request may also review or accept it• High percentage of non reviewed but accepted requests

28 of 30

Questions?

Thank you for your attention.

30 of 30

top related