2026-02-04 Conda Ecosystem Meeting
Zoom link · What time is the meeting in my time zone: 5pm, 2pm
Various parts of the conda ecosystem gather on a regular basis. This meeting brings together all of these sub-communities for a community wide call.
Attendees
- MvN: Marius van Niekerk (@mariusvniekerk), CF (InMarket)
- CHL: Cheng H. Lee (@chenghlee), Anaconda/conda SC/CF core
- KZ: Klaus Zimmermann (@zklaus), Quansight
- Wolf Vollprecht (@wolfv), prefix.dev / conda-forge
- JK, John Kirkham (@jakirkham), NVIDIA & conda-forge/core
- HV, Axel Obermeier (@h-vetinari), conda-forge
- DY, Dan Yeaw (@danyeaw), Anaconda
- JS, Jakov Smolic (@jsmolic), Quansight
- JL, Jannis Leidel (@jezdez), Anaconda/cf
- JRG, Jaime Rodríguez-Guerra (@jaimergp), Quansight, CF/C, C/SC
Introductions
- Welcome OpenKylin folks!
Announcements
- CFEP vote open until 2026-02-06: https://github.com/conda-forge/cfep/pull/64
- Two days left, and only 14 votes out of 25 active core members. Additionally 13 emeritus can vote.
- RFC period open for a bunch of CEPs: https://github.com/conda/ceps/issues/147
- Deadline 2026-02-15. Vote to start on 2026-02-16.
From previous meetings
- (HV) Use our own QEMU instead of Fedora's: discuss approach & next steps
- https://github.com/conda-forge/docker-images/issues/320
- IF: Also need to take care of building the image with that package.
- MvN: Overwrite local image qemu clients with a filesystem mount?
- IF: Can you mount just a file?
- MvN: Recent Docker can.
- MvN: If mounting doesn't work, you can even rebuild the image with the new qemu.
- IF: Also consider using only userspace emulation instead of full filesystem (which require starting Docker with qemu).
- https://github.com/conda-forge/docker-images/issues/320
- (HV) Choice of RISC-V CPU architecture baseline
- https://github.com/conda-forge/ctng-compilers-feedstock/pull/205
- IF: Suggestion to start with rv64gc and GCC 15. Bump to rva23 and GCC 16 and new GLIBC when AlmaLinux does it too (e.g. next year).
- MvN: By then more hardware will be available too.
- JK: And for testing
- IF: OpenKylin has hardware we can use for that.
- JRG: Ubuntu 25.10 requires RV23
- JS: Gentoo has support for rv64gc, and will probably keep that for a while.
- JW: Ideally they want to land rva23 in conda-forge to avoid fragmentation, but the performance gap between rv64gc and rva23 is too big.
- (HV) Move to macOS >=11.0; libcxx v22 will require >=11.0, leaving us little leeway.
- https://github.com/conda-forge/conda-forge.github.io/issues/2467
- https://github.com/conda-forge/conda-forge.github.io/pull/2721
- https://github.com/conda-forge/conda-forge-pinning-feedstock/pull/8162
- Potential alternative: since the split of
libcxx-develfromlibcxx(andclangxx_<target> vNdepending on the matching libc++ headers, i.e.libcxx-devel vN), we would theoretically be able to publish libcxx v22 without raising the deployment target; as long as our default compiler version on osx is <22, the run-export of the C++ compiler would still produce builds compatible with libcxx 21 at runtime. - JK: When are these updates scheduled?
- HV: Coming days or weeks?
- JK: Ok, but let's make an announcement.
- MvN: It's inevitable anyway, right?
New agenda items
- WV: reproducible builds now more serious! https://prefix-dev.github.io/reproducible-builds/v1.html
- WV: Similar efforts by Google: OSS Rebuild. They have attestations
- JK: Related to that, thought of using SLSA attestions?
- WV: Interested in hearing more.
- WV: pixi-gui was released!
- Desktop app that uses Tauri. Manages environments and runs tasks.
- Aiming to provide a simple experience for users that don't like CLIs.
- DY: Why released as source available and not open source?
- WV: To prevent rebranded copies, mostly.
Deferred to next meeting
- JRG (conda-forge): Idea to request OSS sponsorship for blacksmith.sh runners.
- JRG: Bug in CEP 19
- DH: Plan to discontinue
current_repodata.jsonon conda-forge; it can't be used by modern solvers and is slow to generate. - DH: CEPs regarding development policy for the Python implementation of conda. 14-day comment period has elapsed for "use Python API conventions in conda instead of deprecation policy"; alternate new CEP "retract deprecation policy; move policy to conda developer documentation"