Python Wheels

[HugoFara]

Hi, I noticed a big friction point with forefire is the unavailability of a Python wheel. A standard pip install forefire will not work. Looking at the code, I think I can help implement the feature. Would you be interested in a PR?

---

Bonjour, j'apprécie beaucoup le projet forefire, et je pense que faciliter l'installation avec un Wheel Python serait l'outil le plus bénéfique à sa diffusion. Je peux préparer une pull request, est-ce que cela vous intéresserait ?

[antonio-leblanc]

Hi Hugo, thanks for reaching out!

We completely agree that having a pip-installable wheel would be a game changer. The current friction is a known issue, and we would be thrilled to receive a PR for this!

However, we want to be transparent: packaging this won't be trivial. Our setup.py currently assumes the main C++ library (libforefireL) is already built, rather than orchestrating the CMake build itself (which standard wheel builders usually expect, e.g., via scikit-build-core).

Moreover, our dependency on NetCDF (and netcdf_c++4) makes standard cibuildwheel pipelines tricky, as those libraries need to be present in the build environment and properly vendored (via auditwheel/delocate) into the final package.

There's also an MPI axis to consider: MPI support is auto-detected and conditionally compiled in, so the "same" wheel could end up with different capabilities depending on the build environment — something @filippi flagged before in #102.

If you have experience with C++ Python packaging and want to tackle this challenge, you have our full support! Let us know if you need help refactoring the setup.py or CMakeLists.txt to make it happen.

[HugoFara]

Thanks @antonio-leblanc I'm giving it a try! I'll let you know if some features need special adaptation and can't go in the wheel.

For the package name, which do you prefer between forefire and pyforefire? It seems pyforefire is used throughout the codebase, but people often refer to pip install forefire. Both are available on PyPI.

[antonio-leblanc]

Great! Thanks a lot @HugoFara
Concerning the name @filippi is boss to decide the name in pipy.. He is travelling might not answer this week

So I sugget to stay with pyforefire for now since it is already in the codebase

I agree that it might be easier to just use forefire, we can discuss with him when he comes back and probably since both names are available we can change it later!

merci!

[filippi]

Hi yes forefire good Thanks ! ******************************************** Jean-Baptiste Filippi Chargé de Recherches SPE - UMR 6134 CNRS Université de Corse Quartier grossetti BP 52, 20250 Corte Tel. +33 495 450 158 ****************************************** Le mar. 28 juil. 2026, 17:28, Antonio Leblanc ***@***.***> a écrit :

…

*antonio-leblanc* left a comment (forefireAPI/forefire#149) <#149 (comment)> Great! Thanks a lot @HugoFara <https://github.com/HugoFara> Concerning the name @filippi <https://github.com/filippi> is boss to decide the name in pipy.. He is travelling might not answer this week So I sugget to stay with pyforefire for now since it is already in the codebase I agree that it might be easier to just use forefire, we can discuss with him when he comes back and probably since both names are available we can change it later! merci! — Reply to this email directly, view it on GitHub <#149?email_source=notifications&email_token=AAUZ7CUCHKG7WW3CNESVICD5HDBCLA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTKMJQGYYTMMJQHAY2M4TFMFZW63VHNVSW45DJN5XKKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-5106161081>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAUZ7CV36XJINFMNNBUSNLD5HDBCLAVCNFSNUABEKJSXA33TNF2G64TZHMYTOMRUHE3TKMJ3JFZXG5LFHM2DSNZXHA2DCOBWHGQXMAQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>