2019-10-02 conda-forge core meeting¶
Pre-arranged stuff to talk about. Email core team on Monday of the week of the dev meeting to ask for things to talk about.
How do we get greater maitainer participation?
External users find it difficult to get involved
Community ownership model improvements
Have a follow-on conversation with PyTorch maintainers, conda-forge and Anaconda?
(Filipe) NumFOCUS grants and more transperancy/communication with expenses
Write up plan that we discussed on the call and make it public
Create finances subteam to approve disbursement of funds
Numfocus workflow is:
someone submits things through rocket
They update spreadsheet
Someone from conda-forge core needs to approve the new spend.
Set up notifications for your user so that approvers are aware of when changes are made.
(Filipe) CZI proposal (deadline December 2019)
link to bioconda issue
follow up with Bjorn
(CJ) CFEP-10 for assigning standings to feedstocks
Comment in PR that only 3 Version PRs will be issued by bot
What are we supposed to do with conda packages for feedstocks that are no longer being actively maintained
Epoch out packages archived feedstocks?
Package based messaging?
one user perspective: would prefer correct and/or up-to-date versus “works for the thing i did one year ago”
(CJ) Testing story, encourage people to add full test suites?
Mixed responce, table for later
(John) Update on GPU packaging
Made docker images, based on NVIDIA CUDA images
(CJ) NumFOCUS sprint form
(CJ) Problems with Appveyor downgrade to free?
(Marius, Isuru) Azure migration update?
Do we want to roll out https://github.com/conda-forge/conda-smithy/issues/1131 to all feedstocks?
(Isuru) Update to clang 8.0.1 and gfortran 7.3.0 on OSX
No rebuild necessary. (Rebuild for gfortran 7.3.0 is already done. We build with 7.3.0 and 4.8.5)
This should be pain free except for some packages that do -Wall -Werror.
We get better compilers.
We don’t have to rely on defaults as we build these in conda-forge CI.
We get a linker that supports .tbds . (new linker might work with older compiler, but not sure about packages compiled with -flto)
How do we start getting traction on these?
Your agenda items¶
jeremy howard’s criticisms and discussion¶
It’s still difficult to get involved
Ralf, as a numpy / scipy maintainer, still struggles to understand how to get involved.
External project maintainers plus maintainers of recipe – so-and-so already owns it how can i start contributing to the package.
the average user doesn’t want to think about glibc. it’s an implementation detail that the average packager doesn’t want to think about. Most people are happy to follow these guidelines
Check in on previous action items¶
Copy previous action items from last meeting agenda.
[ ] (Eric) Email out to see if we can get more engagement on regular dev meetings.
[ ] (Sophia) Set up meeting to talk about Condafile stuff.
[ ] (CJ) write blog post on bot work and compiler migration
[ ] (Filipe) Ping Bjorn to try and get Debian community more involved with conda-forge
[ ] (Eric) Scheduling Anaconda <-> conda-forge sync on anaconda.org requirements gathering
[ ] (Anthony) Reach out to NumFocus to figure out legal ramifications of not including licenses in files.
2 meetings ago¶
[ ] (Filipe) Progress on setting up a self-hosted blog
Medium seems to be generally downvoted
Host your own blog seems to be generally positive
Can we mimic what Dask does for their dev blog? https://github.com/dask/dask-blog
[ ] (Needs Owner) Searching on our website is a little difficult
[ ] (Filipe, Anthony) Improve docs on releasing conda-smithy. Currently says “use rever”. Would be good to fill out more info there for those of us that are unfamiliar with rever.
3 meetings ago¶
New Action items¶
Copy new action items to next meetings agenda so we can check in. Make sure each action item is assigned to someone or it will likely not get done.