Skip to content

Conversation

@jdblischak jdblischak requested a review from a team as a code owner January 13, 2026 15:24
@conda-forge-admin
Copy link
Contributor

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

  • ℹ️ The recipe is not parsable by parser conda-recipe-manager. The recipe can only be automatically migrated to the new v1 format if it is parseable by conda-recipe-manager.

This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/20962260311. Examine the logs at this URL for more detail.

Copy link
Member

@jjerphan jjerphan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you, @jdblischak.

LGTM (I did not see #7856 initally).

I left comments and approval on feedstocks' PRs, I let you proceed with merging them first.

@jdblischak
Copy link
Member Author

I left comments and approval on feedstocks' PRs, I let you proceed with merging them first.

This time I was going to try and combine the version bumps and the migration PRs

@h-vetinari h-vetinari merged commit 53e94b1 into conda-forge:main Jan 13, 2026
3 checks passed
@jjerphan
Copy link
Member

Hum, mustn't we first have merged all the PRs of feedstocks which have been linked? :)

@h-vetinari
Copy link
Member

Hum, mustn't we first have merged all the PRs of feedstocks which have been linked? :)

Nope, because of exclude_pinned_pkgs: false, those feedstocks will get a migration PR themselves; also you had 👍'd the comment how @jdblischak is planning to combine the migration (PRs) with the version bumps.

@jjerphan
Copy link
Member

jjerphan commented Jan 14, 2026

Nevermind, I misunderstood #8132 (comment) (I thought it meant waiting for all feedstocks to have their version bumped before starting the migration, and I did not know that starting the migration before was possible).

@jdblischak
Copy link
Member Author

As I see it, there are two main strategies for pushing through the updates and migrations for these azure feedstocks:

  1. Merge each update PR as it comes. AFAIU as long as it compiles, it is fine (even if the upstream devs have bumped the minimum required version in the vcpkg.json). Then the combined migration PR needs to designate all the auto-opened single migration PRs to close. I see two main downsides to this approach: 1) it won't work if indeed one of the packages can't be built without updated dependencies, 2) it unnecessarily builds and uploads what are essentially unusable conda binaries (since they are pinned to the previous versions of all its dependencies). This is what I did last time in Migrate azure_core_cpp 1.16.1 and latest storage packages #7856

  2. Delay merging each update PR. Instead allow the combined migration to send migration PRs to each feedstock (because we set exclude_pinned_pkgs: false), and then combine the bot update PR with the bot migration PR. I like this approach because it is cleaner (ie uploads fewer conda binaries). However, it is admittedly more work. Combining the two bot PRs manually is tedious since they both bump the build number and rerender. Depending on how it goes this time, I may decide to switch back to the other strategy to save time.

@jdblischak jdblischak deleted the azure-storage-2026-01 branch January 14, 2026 15:57
@jdblischak
Copy link
Member Author

Depending on how it goes this time, I may decide to switch back to the other strategy to save time.

So far it is off to a bad start. The migration is blocked because the bot can't bot can't solve the environment:

linux_64_: Cannot solve the request because of: No candidates were found for azure-storage-common-cpp 12.12.0.*.

At minimum I'll need to merge the azure-storage-common-cpp 12.12.0 update PR first. Then we'll see if the migration can continue.

@h-vetinari
Copy link
Member

Yeah, the migrator needs at least one library to have been successfully built to solve a child feedstock.

@jdblischak
Copy link
Member Author

While I like that my current strategy avoids uploading unnecessary conda binaries, everything else about it is decidedly worse. Combining the version bump and migration PRs requires more manual intervention, we still have to merge the first package in the dependency network (in this case it was storage-common but would typically be core), and individual PRs to update the pins are still opened (and thus will need to be manually closed).

So going forward, as long as the CI passes, I think we should follow the strategy of #7856: merge all the version bump PRs, create a combined migration PR, list all the independent version bump PRs to be auto-closed.

@jdblischak
Copy link
Member Author

as long as the CI passes

Note that for this migration, the datalake 12.14.0 version bump PR failed without its updated dependencies, so it would have had to have waited for the combined migration regardless (conda-forge/azure-storage-files-datalake-cpp-feedstock#24 (comment), conda-forge/azure-storage-files-datalake-cpp-feedstock#25)

@h-vetinari
Copy link
Member

@jdblischak, I think the synthesis here is that the optimum is somewhere in the middle, and may not be the same from migration to migration. Personally, I think it'd be viable to merge the version updates that pass, and leave the ones open that don't, before starting the migration (for example).

@jdblischak
Copy link
Member Author

Quick status update on the azure storage migration:

@jdblischak jdblischak mentioned this pull request Jan 19, 2026
2 tasks
@jdblischak
Copy link
Member Author

Status updates:

Thanks @h-vetinari and @jjerphan!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants