Skip to content
This repository has been archived by the owner on Aug 22, 2024. It is now read-only.

duplicate feedstock #2

Open
beckermr opened this issue Jun 8, 2021 · 17 comments
Open

duplicate feedstock #2

beckermr opened this issue Jun 8, 2021 · 17 comments

Comments

@beckermr
Copy link
Member

beckermr commented Jun 8, 2021

This feedstock is a duplicate of https://github.com/conda-forge/python-build-feedstock. I was told by @isuruf that we did not want build as the name because it was too generic. I am going to archive this feedstock.

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

cc @conda-forge/build

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

cc @conda-forge/core

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

We're going to wait one week and then archive this one.

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

I got a thumbs up from a maintainer on the staged-recipes PR: conda-forge/staged-recipes#14233

Going to go ahead and archive now.

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

Moving this note over from staged recipes:

@conda-forge/build if we mark all of the build packages a broken, will that break too much of your stuff using the build conda package?

FYI, once a package is marked broken it will no longer be installable in new environments and so CI jobs and other envs that depend on being able to install build will fail.

cc @isuruf

xref: conda-forge/staged-recipes#14233

@chrisburr
Copy link
Member

I think there are sufficiently few downloads that the permanent harm of conda install build giving an unmaintained version is worse that temporarily breaking one or two workflows (if there even are any).

@beckermr
Copy link
Member Author

beckermr commented Jun 8, 2021

Let's give the maintainers a little time to respond.

@jasongrout
Copy link

I'm fine with removing this in favor of python-build. The other two maintainers are build maintainers, so I'd love to hear from one of them too.

Would we automatically submit prs to any package that depends on this package to move it over to the python-build package?

@jasongrout
Copy link

Can the three of us be added as maintainers of the other package as well?

@FFY00
Copy link
Member

FFY00 commented Jun 8, 2021

I have a preference towards this feedstock as I would expect the package to have the same name in conda and pip, but it's not really my call.

I think there are sufficiently few downloads that the permanent harm of conda install build giving an unmaintained version is worse that temporarily breaking one or two workflows (if there even are any).

Isn't there a way to deal with these situations? It's not the first time I run into such issues with conda.

I also struggle a bit to understand the point of the name being too generic if this feedstock will still be reachable. It's not like the users will see an error, the only thing it helps is people being slightly less confused in documented commands.

@isuruf
Copy link
Member

isuruf commented Jun 9, 2021

Would we automatically submit prs to any package that depends on this package to move it over to the python-build package?

Yes, we can make automatic PRs.

Can the three of us be added as maintainers of the other package as well?

Yes, of course.

I also struggle a bit to understand the point of the name being too generic if this feedstock will still be reachable. It's not like the users will see an error, the only thing it helps is people being slightly less confused in documented commands.

One proposal was to not maintain this feedstock, but keep the existing packages as they are. This would make conda install build succeed, but give an old version. We wanted to know if there would be an objection to remove build packages so that conda install build will error and only conda install python-build will work.

@jakirkham
Copy link
Member

Just checking in here, what conclusion did we reach on next steps?

@jasongrout
Copy link

Good question, I've also wondered what the decision was. At this point, comparing downloads, we have 4589 downloads of python-build v0.7.0 and 11668 downloads of build v0.7.0, so I think the reasoning about breaking things and removing packages is very different now than from when the discussion happened.

@beckermr
Copy link
Member Author

beckermr commented Feb 4, 2022

Yeah. The ship has sailed here. Let's add the build output to the python-build feedstock and move on. We can have smithy warn on depending on it and have the bot fix it.

@jakirkham
Copy link
Member

We could add a post-link script with a message warning that people should move to python-build if that is preferred. If having build presents problems for us, I don't think we should let our hands be tied.

That said, not sure whether we want to do that or not.

@beckermr
Copy link
Member Author

beckermr commented Feb 4, 2022

Yeah we should do that and then stop shipping it.

@zklaus
Copy link

zklaus commented Oct 5, 2022

Little nudge then. I just installed mamba install build just guessing the name, then noticed the missing version 0.8.0 and came here to help with the update. Archiving this feedstock would be good.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

7 participants