conda / ceps

Conda Enhancement Proposals
Creative Commons Zero v1.0 Universal
19 stars 24 forks source link

Serve `run_exports.json` in conda channels #51

Closed jaimergp closed 11 months ago

jaimergp commented 1 year ago

As discussed in https://github.com/conda/conda-index/issues/102 and this conda community meeting.

cc @chenghlee

jaimergp commented 1 year ago

Hi @conda-incubator/steering, any thoughts or feedback here? Thanks 🙏

wolfv commented 1 year ago

I think this is a great idea and will also simplify rattler-build!

I am leaning towards adding an empty run exports for all packages (instead of leaving out those that don't have it) so that one knows that these were considered.

wolfv commented 1 year ago

good point @beckermr! There is also pin_as_run in the conda_build_config.yaml file. I believe that that one actually takes precedence over any run exports. Might be good to mention that in the CEP as well.

jaimergp commented 1 year ago

We should state explicitly if we allow patching and if the values in this file taken precedence over the ones in the artifacts.

Right now conda-build (and I would imagine other implementations) have not been notified of run_exports.json existence, so I'd say that initially it won't have an impact on building packages. The file is there as a convenience for automation infrastructure. Taken that into account, I'd be inclined to say that the file is not going to be patched for now.

If the community decides that's convenient, then we can create another (smaller) CEP to change that. I think I have read feedback from some users (cannot recall who, sorry) who were against run_exports patching.

jaimergp commented 1 year ago

There is also pin_as_run in the conda_build_config.yaml file

TIL! I'll check how it works and add it to the CEP. Thanks Wolf!

jaimergp commented 1 year ago

I am leaning towards adding an empty run exports for all packages (instead of leaving out those that don't have it) so that one knows that these were considered.

@wolfv, my thinking here is that the source of truth for all available packages is still repodata.json. run_exports.json simply extends that with extra fields for some. If we enumerate all packages in run_exports.json then we also have to account for removals with "marked as broken" and so on. Does that sound reasonable?

That said, we could always ignore the differences between both. Listing all files might incur some extra download size but will for sure list everything that was analysed, which should make debugging easier...

The more I think about it, the more partial I am to your proposal 😬

jaimergp commented 1 year ago

Added my current view regarding precedence and patching to the CEP. Happy to revisit, but for now I think the easiest is to consider run_exports.json a mirror for the info already available in the archives. This means it should not be patched or given a special treatment wrt pinning resolution in conda-build tooling.

cc @beckermr @wolfv

jaimergp commented 1 year ago

For the record, I plan to call a vote for this in two weeks (next conda community call), so if you have any comments before the vote, this is the time. Thanks! 🙏

chenghlee commented 1 year ago

Suggested some minor tweaks, but I am in favor of this proposal.

jaimergp commented 1 year ago

Thank you @chenghlee! Added your points in the new commits.

jaimergp commented 1 year ago

@conda-incubator/steering, I would like to call a vote for this CEP.

This vote falls under the "Enhancement Proposal Approval" policy of the conda governance policy, please vote and/or comment on this proposal at your earliest convenience.

It needs 60% of the Steering Council to vote yes to pass.

To vote, please leave yes, no or abstain as comments below.

If you have questions concerning the proposal, you may also leave a comment or code review.

This vote will end on August 3rd, 2023 End of Day, Anywhere on Earth (AoE).

jezdez commented 1 year ago

yes

chenghlee commented 1 year ago

Yes

wolfv commented 1 year ago

yes

marcelotrevisani commented 1 year ago

yes!

jezdez commented 1 year ago

This vote will end on August 3rd, 2023 End of Day, Anywhere on Earth (AoE).

@jaimergp Any reason to expand the vote from 7 to 16 days BTW?

jaimergp commented 1 year ago

@jezdez To be able to invoke the timeout quorum rules and have two in-call reminders 😁

jaimergp commented 12 months ago

Yes.

jaimergp commented 11 months ago

Hello team! The vote concluded last week on Aug 3rd at 23:59 AOE (Aug 4th 11:59 UTC). I have now reviewed the votes and these are the results:

--

I will now update the PR to reflect the status and mint a CEP number. Since the vote for this PR was started after #8, this CEP will be minted the number 11.

jaimergp commented 11 months ago

Sorry, small mistake, this CEP will get number 12, not 11.