Add CFEP for package renames #64

jaimergp · 2025-10-26T19:15:07Z

Comes from conda-forge/admin-requests#1711

beckermr

I added a concept called the feedstock epoch to help us track how feedstocks are reused and to help us easily find all of the artifacts from the previous project(s) built in the feedstock.

What do you think?

cfep-XX.md

beckermr · 2025-10-27T16:22:24Z

Also thank you for writing this!

cfep-XX.md

beckermr · 2025-10-29T16:44:14Z

comments @jaimergp ?

cfep-XX.md

beckermr · 2025-10-31T10:32:25Z

@conda-forge/core comments welcome!

isuruf

I really don't like the extra complication with the feedstock epoch. It's just too much to wrap our heads around.

beckermr · 2025-10-31T14:56:25Z

That's fair. My concern is that overtime we'll loose track of the renames and which versions of the feedstock each artifact came from.

I'm sure all of this can be automated.

Would the automation help assuage your concern or is it deeper than the practical bits of actually making all of the labels etc.?

h-vetinari

Thanks for writing this up! I think there's still some things to be fixed/improved.

cfep-XX.md

h-vetinari · 2025-11-02T04:56:35Z

cfep-XX.md

+When an existing package wants to be renamed to a currently unclaimed name, the process is uncontroversial and straight-forward. Two approaches are possible:
+
+- The new name needs to be registered to the originating feedstock in `feedstock-outputs` (or equivalent). Once approved, a PR in the feedstock adding the new output may proceed normally. The old name MUST be kept as an alias output for backwards compatibility. In this case, the feedstock retains the name of the original name.
+- The original feedstock is archived and the package is re-submitted to `staged-recipes` (or equivalent) under the new name. The old package name MUST be added as an alias output for backwards compatibility, and MUST be registered as a valid output of the new feedstock in `feedstock-outputs`.


I'd be tempted to descope this from the first iteration of the CFEP. IMO feedstock rename handling might need its own CFEP, but I don't want to block progress here - given we have a concrete case that would need the policy - based on a concern that's not relevant to that particular case.

But basically my position is that feedstock renames should ~never happen, and if they do, they:

need to meet a very high bar in terms of benefit¹

need to go through staged-recipes

must force-push the old history (or merge it in with --allow-unrelated-histories)

provide an explanation (maybe in a pinned issue or a special README section) about the feedstock history, and where to find the relevant issues/PRs from before

Footnotes

basically, I want to leave the door open for us to do foundational infra work à la https://github.com/conda-forge/ctng-compilers-feedstock/issues/188, but not make this a common occurrence in conda-forge. ↩

cfep-XX.md

beckermr · 2025-11-03T11:43:58Z

OK @isuruf @h-vetinari @jaimergp This is ready for another look!

cfep-XX.md

h-vetinari

Thanks a lot, this is getting better and better IMO! A few remaining nits & open points

cfep-XX.md

isuruf

What are our thoughts on using epochs in the version? conda metadata spec supports epochs.

h-vetinari · 2025-11-04T04:35:18Z

What are our thoughts on using epochs in the version? conda metadata spec supports epochs.

I've made a suggestion about this in one of the review threads. Happy to hear other people's thoughts on this.

beckermr · 2025-11-04T15:51:48Z

What are our thoughts on using epochs in the version? conda metadata spec supports epochs.

They are not very user friendly, but if folks want to use them for their packages, then I'd let them go ahead personally. I personally don't think we need core to review their usage.

Co-authored-by: h-vetinari <[email protected]>

cfep-XX.md

isuruf · 2025-11-04T17:06:24Z

They are not very user friendly, but if folks want to use them for their packages, then I'd let them go ahead personally. I personally don't think we need core to review their usage.

I meant we could require an epoch bump when a package name is reused.

beckermr · 2025-11-04T20:05:15Z

Ohhhh yeah I'm not in favor of that if we don't need it.

h-vetinari · 2025-11-04T22:50:43Z

Here's another fun puzzler about a package name (ouroboros) with a convoluted history (from conda-forge/ouroboros-gis-feedstock#10).

Time	What	PyPI	conda-forge
~2016	Name reserved on PyPI; never any release	`ouroboros` (https://github.com/beeware/ouroboros)	-
July 2025	Unrelated project plans to take over abandoned name; takes name in conda-forge (PR 1, PR 2)	abandoned	`ouroboros` (https://github.com/corbel-spatial/ouroboros)
Aug. 2025	A third project requests the `ouroboros` name, gets it	`ouroboros` (not yet open-sourced)	`ouroboros` (https://github.com/corbel-spatial/ouroboros)
Oct. 2025	The geospatial ouroboros asks for the original PyPI name, realizes it is too late	`ouroboros` (not yet open-sourced)	`ouroboros` (https://github.com/corbel-spatial/ouroboros)

So now we're in a situation where ouroboros (https://github.com/corbel-spatial/ouroboros) was reserved in conda-forge in anticipation of inheriting the PyPI name, which didn't happen for unforeseeable reasons. There's a single release of ouroboros on conda-forge right now, which is just a wrapper package around ouroboros-gis.

I don't think it's worth trying to come up with rules that cover such convoluted cases, but IMO, we should mark that single ouroboros release as broken (let users move to ouroboros-gis), and then keep the ouroboros name free so that we can match the PyPI project that's now called ouroboros, once it makes a release.

beckermr · 2025-11-04T22:56:47Z

Yeah indeed that makes sense! This why practically every other statement in the CFEP comes down to "the core team can take actions to set things right in weird edge cases." ;p

cfep-XX.md

flferretti · 2025-12-05T10:57:32Z

Hi @conda-forge/core, do you perhaps have any update on this? Thank you in advance :)

Add CFEP for package renames

baa7687

jaimergp requested a review from a team as a code owner October 26, 2025 19:15

This was referenced Oct 26, 2025

Un-archive weave feedstock conda-forge/admin-requests#1711

Open

Question: Can this be used as a drop-in replacement for the regular zlib? conda-forge/zlib-ng-feedstock#10

Open

beckermr requested changes Oct 27, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

beckermr reviewed Oct 29, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

feat: add feedstock epoch labels

1fcdaed

beckermr reviewed Oct 31, 2025

View reviewed changes

cfep-XX.md Show resolved Hide resolved

Update cfep-XX.md

940455d

beckermr approved these changes Oct 31, 2025

View reviewed changes

jaimergp mentioned this pull request Oct 31, 2025

Status Report (2025-10-31) Quansight-Labs/conda-ecosystem-sta-mgmt#91

Closed

isuruf reviewed Oct 31, 2025

View reviewed changes

h-vetinari reviewed Nov 2, 2025

View reviewed changes

Update guidelines for renaming packages

b4c76f1

beckermr reviewed Nov 3, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

Update cfep-XX.md

a47396e

h-vetinari reviewed Nov 4, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

cfep-XX.md Show resolved Hide resolved

cfep-XX.md Outdated Show resolved Hide resolved

h-vetinari reviewed Nov 4, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

isuruf reviewed Nov 4, 2025

View reviewed changes

Apply suggestions from code review

2aaae47

Co-authored-by: h-vetinari <[email protected]>

beckermr reviewed Nov 4, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

beckermr added 2 commits November 4, 2025 09:56

Update cfep-XX.md

3b72c42

Update cfep-XX.md

fafe245

h-vetinari mentioned this pull request Nov 4, 2025

Remove ouroboros package from meta.yaml conda-forge/ouroboros-gis-feedstock#10

Merged

5 tasks

h-vetinari mentioned this pull request Nov 4, 2025

Added initial set of packages belonging to the new eozilla ecosystem. conda-forge/staged-recipes#31256

Merged

10 tasks

jaimergp commented Nov 7, 2025

View reviewed changes

cfep-XX.md Show resolved Hide resolved

jaimergp commented Nov 7, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

h-vetinari mentioned this pull request Nov 9, 2025

Status Report (2025-11-07) Quansight-Labs/conda-ecosystem-sta-mgmt#94

Closed

beckermr reviewed Nov 10, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

Update cfep-XX.md

705ba84

beckermr reviewed Nov 10, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

Update cfep-XX.md

de127f9

beckermr reviewed Nov 13, 2025

View reviewed changes

cfep-XX.md Outdated Show resolved Hide resolved

Apply suggestion from @beckermr

f229a2b

flferretti mentioned this pull request Nov 26, 2025

Consider adding Weave as an additional recipe output conda-forge/wandb-feedstock#167

Open

Uh oh!

Add CFEP for package renames #64

Are you sure you want to change the base?

Add CFEP for package renames #64

Uh oh!

Conversation

jaimergp commented Oct 26, 2025 • edited by beckermr Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

beckermr left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

beckermr commented Oct 27, 2025

Uh oh!

Uh oh!

beckermr commented Oct 29, 2025

Uh oh!

Uh oh!

beckermr commented Oct 31, 2025

Uh oh!

isuruf left a comment

Choose a reason for hiding this comment

Uh oh!

beckermr commented Oct 31, 2025

Uh oh!

h-vetinari left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

h-vetinari Nov 2, 2025

Choose a reason for hiding this comment

Footnotes

Uh oh!

Uh oh!

Uh oh!

beckermr commented Nov 3, 2025

Uh oh!

Uh oh!

h-vetinari left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

isuruf left a comment

Choose a reason for hiding this comment

Uh oh!

h-vetinari commented Nov 4, 2025

Uh oh!

beckermr commented Nov 4, 2025

Uh oh!

Uh oh!

isuruf commented Nov 4, 2025

Uh oh!

beckermr commented Nov 4, 2025

Uh oh!

h-vetinari commented Nov 4, 2025

Uh oh!

beckermr commented Nov 4, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

flferretti commented Dec 5, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

jaimergp commented Oct 26, 2025 •

edited by beckermr

Loading