comphead commented on PR #5782: URL: https://github.com/apache/datafusion-comet/pull/5782#issuecomment-5605375687
Validation of the naming fix on the merged head, from [run 34370771074](https://github.com/apache/datafusion-comet/actions/runs/34370771074). The run now carries four distinctly-named native-lib artifacts, where the same run previously produced three records all called `native-lib-linux` plus one `native-lib-iceberg`: | Artifact ID | Name | Producer | | --- | --- | --- | | 10113859066 | `native-lib-linux` | `pr_build_linux.yml` | | 10113741530 | `native-lib-spark-3.5.9-jdk17` | `spark_sql_test_reusable.yml` (Spark 3.5) | | 10113937366 | `native-lib-spark-4.1.3-jdk17` | `spark_sql_test_reusable.yml` (Spark 4.1) | | 10113960518 | `native-lib-iceberg-1.11.0-spark-4.1.3-jdk17` | `iceberg_spark_test_reusable.yml` (Iceberg 1.11) | The consumer side round-trips: 10 `Download native library` steps against the renamed artifacts have succeeded, 7 in `spark-sql-*` jobs and 3 in the Iceberg 1.11 jobs. The 24 `PR Build (Linux)` matrix downloads exercise the unchanged `native-lib-linux`, which now has a single producer. Checks at this point: 28 success, 7 skipped, 0 failed, the rest still running. `Preflight` passed with the new `Check CI config invariants` step green. As before, a green run does not exercise the retry path itself, so this does not test the forced-`overwrite` behaviour. What it does establish is that no producer shares a name with a sibling any more, so an overwrite can only reach that producer's own record. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
