potiuk opened a new pull request, #65986:
URL: https://github.com/apache/airflow/pull/65986
Adds a hygiene pre-pass to `breeze release-management
generate-providers-metadata`: for each provider, every entry in
`provider.yaml`'s `versions:` list **except the first** is checked against
PyPI. Entries PyPI doesn't recognise are stale (e.g. a release that was
prepared but never published) and get removed from the file in place, so
generated metadata reflects only versions a user can actually install.
The first entry is intentionally skipped — it's the in-progress next release
and is allowed to predate publication. PyPI lookups run in parallel; on network
failure the file is left untouched. The version-line removal is regex-based on
the exact shape produced by `_update_version_in_provider_yaml` in
`provider_documentation.py`, so comments and surrounding entries are preserved.
## Verification
- **Happy path:** ran on the full `apache/airflow` provider set — `Checking
provider.yaml versions[1:] against PyPI for stale entries... All provider.yaml
versions[1:] are present on PyPI.` (no false positives across ~100 providers).
- **Prune path:** injected `9.99.9` as `versions[1]` in
`providers/common/compat/provider.yaml` and re-ran:
```
common.compat: removed unreleased versions from provider.yaml: ['9.99.9']
Removed 1 unreleased version entry from provider.yaml files. Re-reading
provider metadata.
```
After the run, `provider.yaml` was back to its original state (sed-inject
+ auto-prune cancelled out — formatting, indentation, and surrounding entries
preserved).
- **First-entry skip:** putting the same fake version in `versions[0]` was a
no-op (per design — the first entry is the in-progress next release).
---
##### Was generative AI tooling used to co-author this PR?
- [X] Yes — Claude Code (Opus 4.7)
Generated-by: Claude Code (Opus 4.7) following [the
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
