Mining software repository with Pharo (ESUG 2025)

Mining software
repository with Pharo
03 July 2025
Nicolas Hlad
with Benoit Verhaeghe & Kilian Bauvent
ESUG2025 - Gdansk

②
③
④
⑤
2
Outlines
Introduction
① What is Mining Software Repository ?
What are the specificities of Berger-Levrault
regarding MSR ?
How we develop GitProjectHealth?
What can you do with it ? (demo)
What is next ?

What is
Mining Software Repository ?
3
Section ①
(MSR)

4
What is Mining Software Repository ?
Mining Software Repository
Definitions - 2
• Today's collaborative development relies on Git Social Platforms (GSP) [Dabbish et al. 2012]
• they are server implementations of Git with builtin social features
• They contain valuable historical information over a software development
Fig - Discussing Pull Request in GitHub Fig - Commits distribution over time (GitHub)

5
Definitions - 2
• Mining Software Repository (MSR) provides methods and tools to extract data
from these platforms [Hassan 2008].
• Among other, it allows us to:
• Studying the impact of code smells [Steﬀen et al. 2010, Palomba et al. 2014]
• Exploring developers code review [Bacchelli et al. 2013]
• Predicting classes prone to change and defect [Bacchelli et al. 2010]
• Retro-analysing a entire development process [Mockus et al. 2000]

6
Existing tools
• General Mining data
• PyDriller — python tool for mining commit
• Git-Miner — Pharo tool, based on CLI-GitMiner
• Specific Mining Data
• Javapers — java lib mostly for Java file analysis in git repository (leveraging SPOON)
• ModelMine — retrieve UML model from project's artefacts
• Data Storage
• Pandora —focus on long terme storage of Git social platform data
• Software Heritage — Archive of Git repository from GSP
• Computing Metrics
• LinearB — Productions and deployment metrics, data positioning with other
companies

What are the specificities of
Berger-Levrault regarding MSR
7
Section ②

8
What are the specificities of Berger-Levrault regarding MSR ?
Industrial context
A quick word on Berger-levrault
• Berger-Levrault is
• a group of international software editors
• with divers sectors of activity (eduction, health, public administration, etc)
• The group acquired divers software editors over the past 30 years.
• From different countries (France, Spain, Canada, Maroco, etc);
• working with different technology (Java, C#, Typescript, Dart, etc);
• and different Git Social Platform (Gitlab, Bitbucket, Azure Devops, etc).

9
Industrial context
A quick word on Berger-levrault - 2
• We use the project management system Altelissan's Jira to manage:
• tickets (Bug, features, Hotfix, etc)
• SPRINT (Agile development)
• releases (delivering dates, testing software, etc).
Fig - Kanban view of a SPRINT in Jira

10
Industrial context
A quick word on Berger-levrault - 3
• Our Jira and Git Social Platform environment are connected
Fig - Linking Jira Tickets to Commit and Merge activity in GitLab

14
Industrial context
The specificities of Berger-Levrault
Git Social Platform project management systems
How to mine from different Git Social Platforms (GSP) ?
How to implement MSR metrics efficiently ?
How to connect GSP data to Jira efficiently ?
We use Model Driven Engineering with Pharo-Moose

How we develop our solution
with MDE
15
Section ③
The conception of Git Project Health

16
GitProjectHealth - MSR with MDE in Pharo
Our MSR solution
GitProjectHealth
GitProjectHealth (GPH) is framework to extract and analyse data from Git Social
platforms using Model-Driven Engineering (MDE).
tool for General Mining data & Metrics computing
• GitProjectHealth contributions are :
• A unify model for all Git Social Platform
• A framework to build custom metric from the model
• A use of metamodel connector to extend any analysis to other platforms (e.g., Jira)
github.com/moosetechnology/GitProjectHealth

17
Main feature
GitProjectHealth
Key Features:
• Data Extraction & importation:
Extracts data from major social pla1orms.
Imports a model of speciﬁc Git en88es.
• Visualization and Metrics:
Visualizes data and computes metrics.
• Model Connection:
Connects models (e.g., Gitlab and Jira).

18
GitProjectHealth
Git Model
Fig - Simplify Metamodel of a Git Social Platform in GitProjectHealth

19
GitProjectHealth
Git Model
1. Naming decisions
Merge Request vs Pull Request

20
GitProjectHealth
Git Model
2. New relations
Commits link to a User, not author

21
GitProjectHealth
Git Model
3. Concepts at fine granularity
Modeling down to the changed line of code

22
GitProjectHealth
API and Importer
Fig — APIs and GSP importers related to our Git model
Our API are independent
open source projects in
Pharo.
Anyone can access them
via github/Evref-BL

23
Petit titre
GitProjectHealth
Metamodel connection: Jira - Git Model
Git model Jira model
Moose Model
GPH-Jira
model connector
sub-model sub-model
The connector accesses all the
entities and relations of its two
submodes (Git and Jira model)

24
Petit titre
GitProjectHealth
Moose Model
GPH-Jira
model connector
sub-model sub-model
The connector accesses all the
entities and relations of its two
submodes (Git and Jira model)

GPH-Jira
model connector
25
Petit titre
GitProjectHealth
commits
issue
message: "[AV1-5886] fix"
id: "1234567"
aGPHCommit aJiraIssue
timespent: "12 days"
key: "AV1-5886"
Connection by attribute value

Our usage of
GitProjectHealth
27
Section ③
Deploying GPH at Berger-Levrault

28
Petit titre
Using GPH at Berger-Levrault
Computing MSR Metrics
• We build a metric framework within GPH
• They are either Projet or User centric
• For each Metric,
• it loads entities from a time period
( i.e., two dates)
• it calculates the metric over a time windows
(i.e. a Day, a Week, a Month, or a Year).
• 47 metrics are implemented so far.
Fig - UML representation of the Metrics in GitProjectHealth
Fig - Running all metrics in GitProjectHealth from a playground (simplified)

29
Petit titre
Using GPH at Berger-Levrault
Metrics computed every weeks (from 2024)

Conclusion & Perspectives
30
Section ⑤
and end.

31
Petit titre
Conclusion
Recap

32
Petit titre
Perspectives
Future Works
• Addressing limitations
• The diﬃculty of maintaining a global metamodel by investigating the generation of
GSP submetamodels from their OpenAPI
• Discuss the purpose of the measures and consider which measures correlate with
a healthy project.
• Evaluating GPH against existing tools (PyDriller, Git-Miner, etc)
• Evolution
• From GPH model to source code model navigating from repository to Famix model
• Build usable knowledge maps by detecting parts of the repository that have
become unknown to developers.

33
Petit titre
Ressources
Links
GitProjectHealth https://github.com/moosetechnology/GitProjectHealth
Pharo Gitlab API https://github.com/Evref-BL/Gitlab-Pharo-API
Pharo BitBucket API https://github.com/Evref-BL/Bitbucket-Pharo-API
Pharo Jira API https://github.com/Evref-BL/Jira-Pharo-API
Example using GitProjectHealth:
Heatmaps https://github.com/Marpioux/Gitlab-HeatMap

34
Citations
Ressources
Bibliography
[Steﬀen et al. 2010] Steﬀen M. Olbrich, Daniela Cruzes, and Dag I. K. Sjùberg. 2010. Are all code smells harmful? A study of God Classes and Brain
Classes in the evolution of three open source systems. In 26th IEEE International Conference on Software Maintenance (ICSM 2010), September
12-18, 2010, Timisoara, Romania. 1-10.
[Palomba et al. 2014] Fabio Palomba, Gabriele Bavota, Massimiliano Di Penta, Rocco Oliveto, and Andrea De Lucia. 2014. Do They Really Smell
Bad? A Study on Developers’ Perception of Bad Code Smells. In Proc. of the 30th International Conference on Software Maintenance and Evolution.
101-110
[Bacchelli et al. 2013] Alberto Bacchelli and Christian Bird. 2013. Expectations, outcomes, and challenges of modern code review. In Proc. of the
35th International Conference on Software Engineering. 712-721
[Bacchelli et al. 2010] Alberto Bacchelli, Marco D’Ambros, and Michele Lanza. 2010. Are popular classes more defect prone?. In International
Conference on Fundamental Approaches to Software Engineering. Springer, 59-73.
[Mockus et al. 2000] Audris Mockus, Roy T Fielding, and James Herbsleb. 2000. A case study of open source software development: the Apache
server. In Proc. of the 22nd international conference on Software engineering. Acm, 263ś272.
[Hassan 2008] A. E. Hassan, "The road ahead for Mining Software Repositories," 2008 Frontiers of Software Maintenance, Beijing, China, 2008, pp.
48-57, doi: 10.1109/FOSM.2008.4659248.
[Dabbish et al. 2012] Dabbish, L., Stuart, C., Tsay, J., & Herbsleb, J. (2012, February). Social coding in GitHub: transparency and collaboration in an
open software repository. In Proceedings of the ACM 2012 conference on computer supported cooperative work (pp. 1277-1286).

36
Petit titre
Titre de section
Texte du titre
Open-Source Analysis (github)
• Analyzed 30 days of activities from Eclipse,
Microsoft, and MooseTechnology.
• Analyzed ~4457 entities over 30 days
• Visualizations: daily commit distributions, user
activity
Industrial Case at Berger-Levrault (gitlab)
Connected Jira model with Git model to analyze
merge request distributions across different issue
types.
• 27% of Merge Requests linked to bug-related
Jira issues.
A user’s commit activity by day, during the
month of September 2024, on moosetechnology
Issues occurrences in September 2024 for
WeHR

0
8
/
2
6
/
2
0
2
4
0
8
/
2
7
/
2
0
2
4
0
8
/
3
1
/
2
0
2
4
0
8
/
3
0
/
2
0
2
4
0
8
/
2
9
/
2
0
2
4
0
8
/
2
8
/
2
0
2
4
0
9
/
0
1
/
2
0
2
4
0
9
/
0
2
/
2
0
2
4
0
9
/
0
3
/
2
0
2
4
0
9
/
0
4
/
2
0
2
4
0
9
/
0
5
/
2
0
2
4
0
9
/
0
6
/
2
0
2
4
0
9
/
0
7
/
2
0
2
4
0
9
/
0
8
/
2
0
2
4
0
9
/
0
9
/
2
0
2
4
0
9
/
1
0
/
2
0
2
4
0
9
/
1
1
/
2
0
2
4
0
9
/
1
2
/
2
0
2
4
0
9
/
1
3
/
2
0
2
4
0
9
/
1
4
/
2
0
2
4
0
9
/
1
5
/
2
0
2
4
0
9
/
1
6
/
2
0
2
4
0
9
/
1
7
/
2
0
2
4
0
9
/
1
8
/
2
0
2
4
0
9
/
1
9
/
2
0
2
4
0
9
/
2
0
/
2
0
2
4
0
9
/
2
1
/
2
0
2
4
0
9
/
2
2
/
2
0
2
4
0
8
/
2
5
/
2
0
2
4
0
8
/
2
4
/
2
0
2
4
#num
of
commits
0
5
10
15
20
Commit activity on vscode - around September 2024
dates
a user

#num
of
commits
0
8
/
2
6
/
2
0
2
4
0
8
/
2
7
/
2
0
2
4
0
8
/
3
1
/
2
0
2
4
0
8
/
3
0
/
2
0
2
4
0
8
/
2
9
/
2
0
2
4
0
8
/
2
8
/
2
0
2
4
0
9
/
0
1
/
2
0
2
4
0
9
/
0
2
/
2
0
2
4
0
9
/
0
3
/
2
0
2
4
0
9
/
0
4
/
2
0
2
4
0
9
/
0
5
/
2
0
2
4
0
9
/
0
6
/
2
0
2
4
0
9
/
0
7
/
2
0
2
4
0
9
/
0
8
/
2
0
2
4
0
9
/
0
9
/
2
0
2
4
0
9
/
1
0
/
2
0
2
4
0
9
/
1
1
/
2
0
2
4
0
9
/
1
2
/
2
0
2
4
0
9
/
1
3
/
2
0
2
4
0
9
/
1
4
/
2
0
2
4
0
9
/
1
5
/
2
0
2
4
0
9
/
1
6
/
2
0
2
4
0
9
/
1
7
/
2
0
2
4
0
9
/
1
8
/
2
0
2
4
0
9
/
1
9
/
2
0
2
4
0
9
/
2
0
/
2
0
2
4
0
9
/
2
1
/
2
0
2
4
0
9
/
2
2
/
2
0
2
4
dates
0
4
8
12
16
Projects

39
Example
Annexe
Sunburst: last activity on the code base
Fig — Sunburst representation of a developer activity in a project

Mining software repository with Pharo (ESUG 2025)

More Related Content

Similar to Mining software repository with Pharo (ESUG 2025) (20)

More from ESUG (20)

Recently uploaded (20)

Mining software repository with Pharo (ESUG 2025)