Contributing to MESSAGEix development

We welcome contributions to the code base and development of new features for the MESSAGEix framework. This page contains guidelines for making these contributions.

Filing issues for bugs and enhancements

We use Github issues for several purposes:

  • Ask and answer questions about intended behaviour or issues running the framework or related models.

  • Report bugs, i.e. unintended or undocumented behaviour.

  • Request changes to exiting behaviour.

  • Request specific enhancements and new features, both urgent and long-term/low-priority.

  • Discuss and design of other improvements.

Please search through open and closed issues for both the message_ix and ixmp repositories. Review any related issues. Then, if your issue is not found, open a new one.

Contributing code via Github PRs

See the short introduction to the Github flow, which describes a pull request and how it is used. Use online documentation for git, Github, and Python to ensure you are able to complete the process below. Register a Github account, if you do not already have one.

1. Choose a repository

Decide which part of the MESSAGEix software stack is the appropriate location for your code:


Contributions not specific to MESSAGEix model framework, e.g. that could be used for other, non-MESSAGE models.


Java / JDBC backend for ixmp.


Contributions not specific to any particular MESSAGEix model instance. Additions to message_ix should be usable in any MESSAGE-scheme model.

message_data or message_doc

Contributions to the MESSAGE-GLOBIOM family of models, including the global model; and its documentation, respectively.

2. Fork, branch, and open a pull request

Fork the chosen repository to your own Github account. Create a branch with an appropriate name:

  • all-lower-case-with-hyphens.

  • issue/1234 if you are addressing a specific issue.

  • feature/do-something if you are adding a new feature.

Open a PR (e.g. on message_ix) to merge your code into the master branch. The message_ix and ixmp repositories each have a template for the text of the PR, including the minimum requirements:

  • A title and one-sentence summary of the change. This is like the abstract of a publication: it should help a developer/reviewer/user quickly learn what the PR is about.

  • Confirm that unit or integration tests have been added or revised to cover the changed code, and that the tests pass (see below).

  • Confirm that documentation of the API and its usage is added or revised as necessary.

  • Add a line to the file describing the changes (use the same title or one-sentence summary as above) and linking to the PR.


  • Include a longer description of the design, or any changes whose purpose is not clear by inspecting code.

  • Put “WIP:” or the construction sign Unicode character (🚧) at the start of the PR title to indicate “work in progress” while you continue to add commits; or use GitHub’s ‘draft’ pull requests. This is good development practice: it ensures the automatic checks pass as you add to the code on your branch.

3. Ensure checks pass

MESSAGEix has several kinds of automatic, or continuous integration, checks:

  • The CLA Assistant ensures you have signed the Contributor License Agreement (text below). All contributors are required to sign the CLA before any pull request can be reviewed. This ensures that all future users can benefit from your contribution, and that your contributions do not infringe on anyone else’s rights.

  • GitHub Actions is used to run several workflows. These are defined by YAML files in .github/workflows/:


    This workflow runs all Python and R tests; on Linux, macOS, and Windows; and for multiple versions of Python.


    This workflow checks for code style and other details:

    • “Lint with flake8”: checks that Code style is met.

    • “Test package build”: checks that the Python package for upload to PyPI, can be built cleanly and without errors.

    • “Test documentation build”: checks that the documentation can be built without fatal errors.


    These tests run daily at 05:00 UTC. They download a particular package of full-scale, MESSAGEix-GLOBIOM global model scenarios from IIASA servers. Each scenario’s outputs are compared to an expected value.

    anaconda, miniconda

    These workflows check that the package can be installed from conda-forge using Anaconda and Miniconda, respectively, on Windows only.

Resolve any non-passing checks—seeking help if needed.

If your PR updates the documentation, the lint check will confirm that it can be built. However, you should also manually build and view the HTML documentation on your machine to confirm that the generated HTML is as expected, and address any warnings generated by Sphinx during the build phase. See doc/README.rst.

4. Review

Using the GitHub sidebar on your PR, request a review from another MESSAGEix contributor. GitHub suggests reviewers; optionally, contact the IIASA Energy Program to ask who should review your code. Address any comments raised by the reviewer, who will also merge your PR when it is ready.

Other tips

  • If other PRs are merged before yours, a merge conflict may arise and must be addressed to complete the above steps. This means that your PR, and the other PR, both modify the same file(s) in the same location(s), and git cannot automatically determine which changes to use. Learn how to:

    • git merge. This brings all updates from the master branch into your PR branch, giving you a chance to fix conflicts and make a new commit.

    • git rebase. This replays your PR branch commits one-by-one, starting from the tip of the master branch (rather than the original starting commit).

Code style

  • Python:

  • R: follow the style of the existing code base.

  • Jupyter notebooks (.ipynb): see below, under Contributing tutorials.

  • Documentation for ReStructuredText in .rst files, and inline in .gms files:

    • Do not hard-wrap lines.

    • Start each sentence on a new line.

    • Ensure Sphinx does not give warnings about ReST syntax for new or modified documentation.

  • GAMS:

    • Wrap lines at 121 characters, except for inline documentation (see above).

  • Other (file names, CLI, etc.): follow the style of the existing code base.

Versions and releases

  • We use semantic versioning.

  • We keep at least two active milestones on each of the message_ix and ixmp repositories:

    • The next minor version. E.g. if the latest release was 3.5, the next minor release/milestone is 3.6.

    • The next major version. E.g. 4.0.

  • The milestones are closed at the time a new version is released. If a major release (e.g. 4.0) is made without the preceding minor release (e.g. 3.6), both are closed together.

  • Every issue and PR must be assigned to a milestone to record the decision/intent to release it at a certain time.

  • New releases are made by Energy Program staff using the Release procedure, and appear on Github, PyPI, and conda-forge.

Contributing tutorials

Developers and users of the MESSAGEix framework are welcome to contribute tutorials, according to the following guidelines. Per the license and CLA, tutorials will become part of the message_ix test suite and will be publicly available.

Developers must ensure new features (including submodules) are fully documented. This can be done via the API documentation (this site) and, optionally, a tutorial. These have complementary purposes:

  • The API documentation (built using Sphinx and ReadTheDocs) must completely, but succinctly, describe the arguments and behaviour of every class and method in the code.

  • Tutorials serve as structured learning exercises for the classroom or self-study. The intended learning outcome for each tutorial is that students understand how the model framework API may be used for scientific research, and can begin to implement their own models or model changes.

Coding & writing style

  • Tutorials are formatted as Jupyter notebooks in Python or R.

  • Commit ‘bare’ notebooks in git, i.e. without cell output. Notebooks will be run and rendered when the documentation is generated.

  • Add a line to tests/, so that the parametrized test function runs the tutorial (as noted at PR #196).

  • Optionally, use Jupyter notebook slide-show features so that the tutorial can be viewed as a presentation.

  • When relevant, provide links to publications or sources that provide greater detail for the methodology, data, or other packages used.

  • Providing the mathematical formulation in the tutorial itself is optional.

  • Framework specific variables and parameters or functions must be in italic.

  • Relevant figures, tables, or diagrams should be added to the tutorial if these can help users to understand concepts.

    • Place rendered versions of graphics in a directory with the tutorial (see Location below). Use SVG, PNG, JPG, or other web-ready formats.


Generally, a tutorial should have the following elements or sections.

  • Tutorial introduction:

    • The general overview of tutorial.

    • The intended learning outcome.

    • An explanation of which features are covered.

    • Reference and provide links to any tutorials that are interlinked or part of a series.

  • Description of individual steps:

    • A brief explanation of the step.

    • A link to any relevant mathematical documentation.

  • Modeling results and visualizations:

    • Model outputs and post-processing calculations in tutorials should demonstrate usage of the message_ix.reporting module.

    • Plots to depict results should use pyam.

    • Include a brief discussion of insights from the results, in line with the learning objectives.

  • Exercises: include self-test questions, small activities, and exercises at the end of a tutorial so that users (and instructors, if any) can check their learning.


Place notebooks in an appropriate location:


Stand-alone tutorial.


Group of tutorials named “example.” Each notebook’s file name begins with the group name, followed by a name beginning with underscores. The group name can refer to a specific RES shared across multiple tutorials. Some example names include:


<group>_basic.ipynb  # Basic modeling features, e.g.:

<group>_adv.ipynb  # Advanced modeling features, e.g.:

<group>_renewables.ipynb  # Features related to renewable energy, e.g.:

Contributor License Agreement

Summary and scope

It may seem self-evident that contributing to a project distributed under an open-source license is an implicit permission to anyone for using the contributed code. However, a formal Contributor License Agreements (CLA) makes contribution terms explicit and provides the project maintainers a record of your agreement to those terms.

A wide range of terms exist in other CLAs, including waiver of moral rights, consequential damages waiver, as-is disclaimer, etc. For this project, we follow the more bare-boned GitHub CLA, which focuses on the three most important clauses: copyright, patent, and source of contribution.

In short, by signing this Contributor License Agreement, you confirm that:

  1. Anyone can use your contributions anywhere, for free, forever.

  2. Your contributions do not infringe on anyone else’s rights.

Definition of terms

The following terms are used throughout this agreement:

  • You - the person or legal entity including its affiliates asked to accept this agreement. An affiliate is any entity that controls or is controlled by the legal entity, or is under common control with it.

  • Project - the repositories message_ix and ixmp, and any derived repositories, projects, or software/code packages.

  • Contribution - any type of work that is submitted to a Project, including any modifications or additions to existing work.

  • Submitted - conveyed to a Project via a pull request, commit, issue, or any form of electronic, written, or verbal communication with GitHub, contributors or maintainers.

2. Grant of Patent License

Subject to the terms and conditions of this agreement, You grant to the Projects’ maintainers, contributors and users a perpetual, worldwide, unlimited in scope, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer your contributions, in whole or in part, where such license applies only to those patent claims licensable by you that are necessarily infringed by your contribution or by combination of your contribution with the project to which this contribution was submitted.

If any entity institutes patent litigation - including cross-claim or counterclaim in a lawsuit - against You alleging that your contribution or any project it was submitted to constitutes or is responsible for direct or contributory patent infringement, then any patent licenses granted to that entity under this agreement shall terminate as of the date such litigation is filed.

3. Source of Contribution

Your contribution is either your original creation or based upon previous work that, to the best of your knowledge, is covered under an appropriate open source license. You assure that you are legally entitled to submit your contribution and grant the above license, or you have clearly identified the source of the contribution and any license or other restriction (like related patents, trademarks, and license agreements) of which you are personally aware. If your employer(s) or employee(s) have rights to intellectual property that you create, you represent that you have received permission to make the contributions on behalf of that employer/employee, or that your employer/employee has waived such rights for your contributions.

Should the licensor be held responsible for any violation of intellectual property right in relation to your contribution, you shall be fully liable for damages that may arise.

Reference and License

This Contributor License Agreement and the introductory text is adapted from the GitHub Contributor License Agreement, Version 298f3afd updated August 9, 2017. GitHub granted a CC-BY-4.0 License to IIASA to use and modify the text of the CLA.