Towards Better Open-Source Development:

Towards Better
Open-Source
Development:
Improving PyQtGraph’s
Feature-Development
Process
Thesis Presentation
By Aditya Kelekar
BE (IT) Metropolia University of Applied Sciences

-
– Let’s spare a moment to think about
what is happening with a giant open-
source software project….
At a well-known open-source project

Source: Linux Kernel Report 2017, Linux Foundation
Figure 1:
Top companies
contributing to
the Linux kernel,
4.8– 4.13 in 2017
Linux Kernel Contributors

Table of Contents
– 1. What is PyQtGraph and where does it come from?
– 2. Open Source Feature Development: Known Facts
– 3. Analysis of PyQtGraph’s Feature Development Process
– 4. Guidelines for PyQtGraph’s Feature Process
Improvements
– 5. Conclusions

PyQtGraph: A graphic library
Functionalities:
– Basic 2D plotting
– Image display with interactive
lookup tables
– 3D graphics system
– Library of widgets and modules
useful for science/engineering
applications
Source: www.pyqtgraph.orgFigure 2: Histogram drawn with
PyQtGraph

PyQtGraph:
Components & Competitors
Figure 3:
PyQtGraph’s Dependencies
and Other Graphics Libraries
NOTE: Size of shapes is not an
indicator of any metric

Feature Development in Open-Soure Software
– Iterative process with a public repository
– Mailing list, Forum Boards
– Small, frequent changes to code repository
– Few key developers (that is, limited resources)
– Atleast one maintainer

Applying Pirate Metrics to
PyQtGraph Project
Figure 4: The
AARRR! Metrics
for PyQtGraph
Source:
Pirate Metrics: A new
way to measure open
source community
success by Gaby Fachler

To Accept or Not to Accept?
– A dilemma often presenting itself to the maintainer:
– One side:
– Accepting (new) code appeases the feature contributor; (possibly also) other
users
– Other side:
– New code becomes the responsibility of the maintainer

PyQtGraph’s Code Development
– Bug Reports and New Feature Proposals on GitHub Issues, GitHub Pull Request
and PyQtGraph GoogleGroups pages
– Maintainer of the GitHub (and also founder): Luke Campagnola
– 8-10 user queries/feature proposals every month
– 60 percent of user queries/feature proposals are answered
– About 40 ‘listed’ contributors
– All development is voluntary-based
– FAQ for prospective contributors is available

PyQtGraph Google Group Statistics
Figure 5: Data Related to Number of Posts on PyQtGraph ’Google
Group’ Forum site

Analysing the Library Forum Posts
– Only posts where the maintainer had commented were analysed
– Corresponding changes in code in Github were studied
– A list of observations was created
– 3 cases of feature development were studied
– The 3 cases represented different feature development outcomes

A Successful Development Cycle
aa
Figure 6: Timeline
of events for a
typical successful
feature-addition
process.

Case of Unsuccessful Feature
Development
Figure 7:
Timeline of
interactions for
the “New Time
Axis” proposed
feature

Suggested Improvements for Feature
Development Process
– Need for a Collaboration Tool.
(Objective: focus the current development resources towards feature completion)
– A new metric to assign collaboration level for new feature code posts
– Visibility of across GithHub and Google Groups forum
– While feature development in progress: correction list auto-tracking features

Pirate Metrics + Interactions
Component
Figure 8:
Extended
Pirate Metrics
with
Interactions
component

PyQtGraph’s GitHub Pull
Requests Page

Conclusions: Beneficiaries &
Limitiations of Scope
– This study could aid:
• a developer wishing to contribute to the PyQtGraph project code
• maintainer of the PyQtGraph project
• User studying the open-source process
- Limitations:
 Research based only on one open-source library
 Each open-source project may have its own dynamics

References:
– 1. Luke Campagnola. PyQtGraph Project Home page:
http://www.pyqtgraph.org/ [Internet] [cited 24 April 2018]
– 2. Luke Campagnola. PyQtGraph Project Official Documentation page:
http://www.pyqtgraph.org/documentation/installation.html [Internet] [cited 24
April 2018]
– 3. Pirate Metrics: A new way to measure open source community success.
https://opensource.com/business/16/6/pirate-metrics [Internet] [cited 24 April
2018]

Towards Better Open-Source Development:

More Related Content

Similar to Towards Better Open-Source Development: (20)

Recently uploaded (20)

Towards Better Open-Source Development: