Skip to main content

conda-forge core meeting 2024-08-21

Add new agenda items under the Your __new__() agenda items heading

Attendees​

NameInitialsGitHub IDAffiliation
Jaime Rodríguez-GuerraJRGjaimergpQuansight/cf
Filipe FernandesFFocefpafconda-forge
Jannis LeidelJLjezdezAnaconda/cf
John KirkhamJKjakirkhamNVIDIA/cf
Marcel BargullMBmbargullBioconda/cf
Klaus ZimmermannKZzklausQuansight
Cheng H. LeeCHLchenghleeAnaconda/cf
Dasha GurovaDGdashagurovaAnaconda/conda
Matthew R BeckerMRBbeckermrconda-forge

X people total

Standing items​

  • [ ]

From previous meeting(s)​

  • [ ]

Active votes​

  • [ ]

Your __new__() agenda items​

  • (JK) Closing out issue on CentOS 7 / GLIBC 2.17

    • https://github.com/conda-forge/conda-forge.github.io/issues/1436#issuecomment-2263984174
    • (HV) See also questions around removing sysroot hacks
    • MB: bioconda not using stdlib functions yet, so it's a good example of how this breakage can manifest outside conda-forge
    • HV: We should document the need for stdlib functions if people want to use our compilers.
    • MB: We will need to keep track_features around while we "deprecate" the old way of using the compilers. This will give downstream time to adjust to the new way, if they can modify their recipes. For other use cases, we might need to think about the possible implications.
    • IF: Main issues in openmpi etc were related to ldflags not being set correctly. This has been addressed. As a result, not very time-sensitive and we can take our time to get things right.
    • HV: Make announcement about the deprecations / requirements for stdlib.
    • IF: If the user wants to use the compilers locally on their machines, then our compilers should know how to use the system sysroot (instead of conda-forge's). A new issue with the actual problem seems more useful.
    • Action items:
      • Reply to the issue mentioning the bioconda situation
      • Ask for a new issue with the actual problem so we can come up with a better solution
    • JK: If this sysroot issue is not a blocker, can we then close 2.17?
    • MRB: Watch for current_repodata.json hacks, because we are adding some metapackages there for the sysroots. This will not break it, but it will make solves harder for classic. (libmamba does not use current_repodata).
    • MB: Ask Anaconda about conda versions usage, solver usage.
      • Jaime: We want the anonymous user agent stirngs
      • JL: Create issue in conda/infrastructure for follow up.

  • (JK) Travis?

  • (JK) Python 3.13

  • (KZ) python-build vs build issue

    • Brief summary:
      • The builder https://build.pypa.io was first added in the python-build-feedstock in 2020 and then again by error in the build-feedstock in 2021, which was subsequently archived.
      • Now there are both python-build and build packages, but the build packages are horribly outdated
      • People and packagers use build, find it outdated and run around confused until they come upon python-build.
      • We have a migrator hanging around on the status page with all entries 0.
    • Ways forward:
      • Add an alias build to python-build so both names work with current versions?
      • Mark all build packages broken to force people to migrate?
      • Close out migrator/finish it if needed?
    • MRB: where has this happened recently?
      • KZ: I am not aware of any actual, recent issues. I just stumbled into this again because I was looking to make some headway with migrators in general and this one, with 0 everywhere, stuck out and reminded me of the discussion.
    • MRB: we should add an entry to this file: https://github.com/conda-forge/conda-forge-pinning-feedstock/blob/main/recipe/linter_hints/hints.toml to make the linter warn about using build. See PR https://github.com/conda-forge/conda-forge-pinning-feedstock/pull/6313
    • MRB: FWIW, the migrator will repopulate itself if any new feedstocks add a refernce to build and then the bot will send a PR.
    • to do items:

  • (IF) Instructions for downstream users of conda-forge recipes

    • https://github.com/prefix-dev/reproducible-builds/issues/182
    • (CHL) Anaconda will follow up to ensure recipe licenses are properly handled (copied into packages' metadata directory; modified/extended for Anaconda changes; etc.)
    • Follow up with Wolf on prefix
    • (TODO) Document how recipe license file gets copied

  • (DG) Anaconda team planning to update the Channel Cloning infrastructure for anaconda.org

    • CDN mirroring is changing the underlying infrastructure used. This should improve mirroring and up-time.
    • Need to check on conda-static usage (as this wouldn't be supported in new system?)
      • Maybe used for RSS, but that might not work now
      • Not used a lot, but need to update the usage on the conda-forge website
    • CHL: To raise issue on webpage repo explaining planned changes for feedback

Pushed to next meeting​

CFEPs​

  • [ ]