On Sat, Jun 29, 2024 at 9:21 AM Bill Allombert <ballo...@debian.org> wrote:
> Dear Debian science, > > The jupyter-notebook packages are in bad shape, > there are 4 RC bugs and the last maintainer upload was nearly 2 years ago, > the version in unstable does not start, there are new usptreamm versions > etc. > Big systems like Jupyter are a problem for distro maintainers. They often depend on packages not used elsewhere and are slow to support current Python versions. Distro versions of big libraries (HDF, NetCDF4, GDAL) are often built without some specialized configuration options that are needed by some Python tools. Distro packagers are rarely heavy users of Anaconda, so are not able to verify that all use-cases work properly. The instructions from "https://jupyter.org/install"; say: This page uses instructions with pip <https://pip.pypa.io/en/stable/>, the recommended installation tool for Python <https://packaging.python.org/en/latest/guides/tool-recommendations/#installation-tool-recommendations> . If you require *environment management* as opposed to just installation, look into conda <https://docs.conda.io/>, mamba <https://mamba.readthedocs.io/>, pipenv <https://pipenv.pypa.io/>, and Homebrew <https://brew.sh>. Many Jupyter users have colleagues in other institutions so it makes sense to work with conda/mamba using anaconda.org packages so a distributed, cross-platform, group has access to the same versions. There are similar considerations for Sage Math and Julia. There are some other large systems (ESA SNAP, NASA SeaDAS) where all users share a common code base and the systems provide their own set of libraries to avoid the problems of differences across linux distributions. > > Maybe we can do something about it. > Focus on the core elements (display, filesystems, networking, security) so Debian is a reliable host for large, well-maintained, 3rd party systems, including anaconda.org. > > Cheers, > -- > Bill. <ballo...@debian.org> > > Imagine a large red swirl here. > > -- George N. White III
