--- # System prepended metadata title: Fatiando Development Calls 2026 --- # Fatiando Development Calls 2026 📱 **Join the video call:** https://meet.jit.si/fatiando-a-terra 📅 **Shared calendar:** https://www.fatiando.org/calendar > **IMPORTANT**: Everyone is required to follow our > [Code of Conduct](https://github.com/fatiando/community/blob/main/CODE_OF_CONDUCT.md) > when participating in the Fatiando community. Please review it carefully. Development calls are **open to everyone**! Here we discuss project development, socialize, and do some live coding from time to time. **Everyone is encouraged to participate and edit the notes below.** -------------------------------------------------------------- ## Next meeting ## 2026-05-06 ### Participants * Matt * Leo * Arthur ### Notes * Mostly just chatting * State of xrft in Harmonica * Review of the draft AI policy https://github.com/fatiando/community/pull/177 ## 2026-04-22 ### Participants * Santi * ### Notes Policy on Generative AI usage: * Started a draft in https://github.com/fatiando/community/pull/177 * Discuss policy on usage of genAI tools in code review. * Santi would like to forbid using genAI tools to write the code review. But there are a few acceptable uses: writing a snippet to show an example, add extra tests, fix grammar. Basically, all accepted uses for developing code and writing documentation. * We should improve that sentence. * The text might be too long. Should we trim it? Potential issue with xrft and next Harmonica release: * https://github.com/fatiando/harmonica/issues/646 * Santi would like to release Harmonica 0.8.0 this or next week. * Should we remove the constrain on xrft as dependency for now? It's very unlikely that the issue will get solved upstream during this week. * Should we implement our own ffts on dataarrays instead? * Maybe we can remove the version constrain in the conda-forge feedstock instead? ## 2026-04-08 ### Participants * Santi * Leo * Arthur ### Notes * Reviewing the Euler Inversion PR in Harmonica: https://github.com/fatiando/harmonica/pull/660 * Arthur wants to start coding in Fatiando. Wil probably start by adding spherical kernels for Choclo. Maybe finish Lu's PR: https://github.com/fatiando/choclo/pull/57 ## 2026-03-25 ### Participants * Santi * Leo ### Notes * Verde v1.9.0 was released! * Now we can get rid of pinning of sklearn in Harmonica's CI * Idea for 2.0: in KNeighbors.predict weight the reduction by distance. That will make the predictions less dependent on the value of k. * Boule got some love * ruff, no codecov, python 3.13, src layout * breaking change for new version: coordinates are now taken as a tuple. Specify what type of coordinates through the coordinate_system argument * we will need to bump minimum version of boule in Harmonica ## 2026-03-11 ### Participants * Santi * Leo * Mariana ### Notes - Thinking about applying for a fiscal host: - Open Collective: https://documentation.opencollective.com/ - Choosing fiscal hosts: - https://opencollective.com/opensource - https://opencollective.com/numfocus (things to consider: https://pirsquared.org/blog/numfocus-concerns.html#NF-europe) - https://opencollective.com/europe - Do they need us to be based there? - Example of coops: https://coop.planetary-research.org/ - AI usage in Fatiando: https://github.com/fatiando/community/issues/171 - No PRs from agents. - Checklist item on PR template about AI use. - Suspected AI generated PRs can be closed without lengthy justification. - Contributions are for us to teach and train new contributors and not bots. - Put summary of the policy in all the READMEs - State preference about human-written code - Copyright issues: - License infringements without noticing - US court saying that genAI products are not copyrightable - Hardly going to be enforced - Trust people to do the right thing. - Talk about our position and how we envision Fatiando. - Say we prefer no AI since we're about getting people together to work on things - Add examples of acceptable uses for genAI - Translation to English if you are not a native speaker - Use AI to review your code - Add section to general CONTRIBUTING.md and add sections with links to project READMEs and CONTRIBUTING files. ## 2026-02-25 ### Participants * Dom * Tizian * Santi * Pascal ### Notes Tizian, Pascal and Dom shared their progress on `pooch-doi`, the new package that will handle fetching files from DOI-based services, like DataVerse, Figshare and Zenodo. Naming packages, ideas: - `pooch-doi` - `pooch-zenodo` - `pooch-figshare` - `pooch-dataverse` Potentially, we could have more repositories for other data repositories. Where do the new pooch-related packages live? We could create repos in the fatiando organziation for them, and appoint Dom, and anyone else that would like to step in, as maintainers. ## 2026-01-28 ### Participants * Santi * Collin Cronkite-Ratcliff ### Notes > :mega: **Announcement** :mega: > In case you missed it: Matt Tankersely has joined as Harmonica maintainer! Welcome! :raised_hands: Harmonica: - Created an `environment.yml` file used to build docs in GitHub Actions: https://github.com/fatiando/harmonica/pull/626 - Instead of collecting requirements from the `env` files with `dependente` for two main reasons: those packages have PyPI names (we needed to rename a few to get the right conda package, like `build` into `python-build`), and with this file we can constrain particular versions of the optional packages if anything stop working upstream. - Had to constrain `sklearn` version to `<1.8.0` since `1.8.0` is making Verde to error out because of missing attribute. See https://github.com/fatiando/harmonica/pull/628. - Fixed docs failing due to incompatibilty with Pandas 3.0.0 - The `DataFrame.ravel()` method got deprecated, so I patched the equivalent sources to use `np.ravel()` on the input data array instead. This way we also ensure that the ravelled data is always a Numpy array. Verde: - The `DummyEstimator` is failing with `sklearn==1.8.0` due to a missing attribute `__sklearn_tags__`. See https://github.com/fatiando/verde/issues/530 - Possible solution: make it a child of `sklearn.base.BaseEstimator`. See https://github.com/fatiando/verde/pull/531 Pooch: - Merged a few PRs - Better error messages when requests fail. - Support for Python 3.14 - Patch to bypass (at one extent) Zenodo's rate limits. Pass a default Pooch's User-Agent when making requests. - Still some issues with how DOI downloaders are designed. Hard to standardize all requests. First try in this PR: https://github.com/fatiando/pooch/pull/514 - Planning to release this/next week, so users can make use of these changes. USGS: - Collin mentioned that they'll work on a USGS open-source project to develop a workflow for processing potential field data. - Mostly interested in gravity for now. - They will use Fatiando for the underlying computations. - Open to contribute back if there's some methods that would be nice to have in Fatiando.