dabla opened a new pull request, #74673: URL: https://github.com/apache/airflow/pull/74673
Providers that make an operator durable through `ResumableJobMixin` (AIP-103) still support Airflow versions that do not have it. So far, each of them carries its own fallback for below 3.3. There are seven identical copies, each with the same sentinel and warning helper: - Glue - Redshift Data - Databricks - BigQuery - Snowflake - Livy - Spark Submit Each copy also has its own tests, and the next operator to become durable would add an eighth. `ResumableJobMixin` is now available from `airflow.providers.common.compat.sdk`, in the same way as `ChainRetryPolicy`: - **3.3 and later:** you get the SDK class. - **Below 3.3:** you get a stand-in that submits a new job on every try, which is what the operator did before it became durable. The operator keeps working on older Airflow, just without resuming. - **Warning:** setting `durable` explicitly on an old version warns that it has no effect. Leaving it at its default does not warn. The seven operators now import it from there, and their own copies and tests are gone. Two things follow: - **One copy instead of one per provider (DRY).** A fix to the fallback lands once. - **Making an operator durable gets simpler.** A provider imports the mixin from `common.compat`, implements the six methods, and calls `execute_resumable`. It no longer has to write and test its own fallback for older Airflow. The Microsoft Azure PowerBI and MSGraph operators are the next to become durable, in a PR stacked on this one. --- ##### Was generative AI tooling used to co-author this PR? - [X] Yes (please specify the tool below) Generated-by: Claude Code (Opus 5.5) 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]
