Skip to content

Latest commit

 

History

History
111 lines (92 loc) · 5.44 KB

ROADMAP.md

File metadata and controls

111 lines (92 loc) · 5.44 KB
Raw

Roadmap

GrimoireLab was born with the purpose of offering a free, libre, and open source data platform for analytics and insights about software development processes.

In GrimoireLab, we develop tools that help developers and managers to make better decisions for their daily work. Our tools are also designed for the research and academic community.

We believe in open source and in empirical analysis. We think there's a lack of these tools in the data analysis ecosystem. Our goals are:

  • To be the reference platform for retrieving and analyzing data about software development.
  • To build an extensible platform, so other tools can built their solutions on top of it.
  • To improve knowledge and innovation in the field of software engineering.

This document enumerates the steps we will take to achieve these goals and how the platform should evolve in that regard. For more information about how we manage the ROADMAP, please check the Roadmap section in the GOVERNANCE file.

Context

Over the years, we've built a platform that analyzes data from more than 30 data sources, producing more than 150 metrics about activity, performance, and community in software development.

Our platform has been used by numerous companies and by some of the most important open source projects that needed to better understand their software development processes.

The platform has also been used as the base for building other applications on top of it, such as Bitergia Analytics, OSS Compass, Cauldron, or Mystic,

Despite this relative success, GrimoireLab hasn't reached a stable version (1.0) yet. We understand that our software has several issues that prevent our users and developers from getting the most out of it. We have identified the following challenges:

  • Installation and configuration: The software is complicated to install and require a significant amount of time and knowledge to set it running. Some parts of the documentation are also outdated and/or inaccurate.
  • Maintenance: The maintenance of the platform requires several manual steps. Monitoring is also difficult because the platform doesn't provide clear information about what tasks are running and what errors have been found.
  • Scalability and performance: The current version of the platform needs of several instances for analyzing more than 5000 data sources. For example, for a project with around 3500 high-activity repositories that retrieves data from GitHub (commits, issues and pull requests), the platform needs 3 days to start analyzing new data.
  • Integration: The current architecture doesn't make life easy for developers who want to integrate GrimoireLab with their tools. For example, the lack of a well defined data model prevents data analysis tools from generating reusable new metrics. There's no API to add or remove data sources to analyze or to get the metrics from a particular case.
  • Relation with other CHAOSS tools: There's some overlap in the CHAOSS software tools that makes us feel that we're reinventing the wheel. Also, people that reach our project through CHAOSS often don't understand the difference between each tool. We think we can be the reference for data retrieval and initial analysis, so other tools can be build on top of GrimoireLab.

Some of these problems were also mentioned by our users on the 2023 CHAOSS Challenges Survey.

Objectives

Based on these challenges, we have defined a set of goals:

  • Reduce set up time: The installation and configuration of the platform shouldn't take more than 15 minutes for someone with intermediate computing skills and knowledge about software data analysis.
  • Cut manual maintenance to zero: For usual common cases, such as: updating or rebooting the platform, managing retrieval or analysis tasks, and regenerating data, the manual interaction with GrimoireLab should tend to zero.
  • Increase the throughput: The platform should be able to analyze 3500 high-activity GitHub repositories in less than one day.
  • Be the CHAOSS reference tool for data retrieval: Our tool will produce data that will be consumed by other tools for their analysis in the CHAOSS environment (i.e. augur).

Themes

Tasks will be classified under the following themes. Themes are defined to accomplish the objectives and solve the problems described above.

  • Configuration: To make easier how to set up and run the platform.
  • Data processing: To scale the platform for analyzing thousands of repositories. We have defined these sub-themes:
    • Events: To make an event-driven platform that will improve how data is processed.
    • Scheduling: To define how tasks can run without timeouts and penalties.
  • Data access: To solve the integration problems and how data is consumed. The sub-themes are:
    • Data model: For a better representation and description of the data we store and produce.
    • API: To define different levels of access to data and to support the integration with other systems.

Check the Roadmap section in our GOVERNANCE document for more info about themes.