conda-forge core meeting 2025-04-02
Add new agenda items under the "Your __new__() agenda items" heading
Attendees
| Name | Initials | GitHub ID | Affiliation |
|---|---|---|---|
| marcoesters | ME | marcoesters | Anaconda |
| Dawn Wages | DW | dawnwages | Anaconda |
| Jaime Rodríguez-Guerra | JRG | jaimergp | Quansight/cf |
| Daniel Ching | DJC | carterbox | NVIDIA/cf |
| Jannis Leidel | JL | jezdez | Anaconda/cf |
| matthew becker | mrb | beckermr | cf |
X people total
Standing items
- [ ]
From previous meeting(s)
- [ ]
Active votes
- [ ]
Introductions
- (DW) At Anaconda for a few weeks. Previously Django, Microsoft and more. Will interface with the community a lot. Coming in as the Director of Community and Developer relations at Anaconda.
Your __new__() agenda items
- (HV) 10th anniversary coming up on April 11th -- let's celebrate that visibly (even if it's only a blog post).
- Jaime will draft something
- (HV) Closing the Numpy 2.0 migration? https://github.com/conda-forge/conda-forge-pinning-feedstock/issues/7222
- If we do close it, post an announcement.
- MRB: ~30 unsent PRs because they are awaiting parents. We could force-send them so at least they have a notice about the situation. This probably needs a little PR in the bots to by pass the 'awaiting-parents' checks.
- (HV) BLAS situation: MKL 2025 has broken something in the way it does its "single dynamic library", which we're using to make the blas-metapackage work. Tried patching things, but not successful. This is pretty painful because it pushes the solver into weird places and should be fixed ASAP IMO. Help most welcome.
- (DJC) Will try to reach out to Intel folks directly through professional network
- (HV) How to deal with macOS lower bound going forward: https://github.com/conda-forge/conda-forge.github.io/issues/2467
- Favor changes in pinnings and smithy
- DJC: Why not follow Apple's policies?
- IF: Too aggressive; we would need to update very often and leave a lot of packages behind. Even Homebrew stays a bit behind on purpose.
- DJC: We should have both a lower bound policy (e.g. 7 years old version) and a mechanism for migrating dependency trees for groups of packages which need a newer development toolchain
- IF: That might be controversial given the multiple stakeholders in conda-forge. Agree it's necessary but expect long discussions.
- (ME) Introducing a composite action to create
constructor-based installers (Miniconda, Miniforge) - (JRG) Enforcing emeritus provisions
- (IF) Duplicate RPATHs disallowed in macOS 15.4+. Need to teach conda to remove if existing
- already showing up in numpy: https://github.com/conda-forge/numpy-feedstock/issues/347
- Action item: raise issue in conda-build
- Thread at: https://conda-forge.zulipchat.com/#narrow/channel/457337-general/topic/Multiple.20LD_RPATH.20in.20libgfortran
- (JRG) New features in conda-metadata-app:
- Linked dependency lists (check option on the sidebar)
- Report artifact size
- Remember subdir choices while browsing
- (HV) [stretch; sort after current affairs] Discussion about rebuilding the world https://github.com/conda-forge/conda-forge.github.io/issues/2295 (VS 18.0 has appeared on the horizon; might break ABI)
Pushed to next meeting
- [ ]
CFEPs
- [ ]