Once we get support for merge-append https://github.com/apache/iceberg-python/pull/363 operations, and also compaction in the future, it would be good to have rolling-manifest writers that open a new file once the manifest gets to a certain size (both bytes and number of entries). This makes sure that Iceberg tables have decent parallelization when doing query planning.
Hi, I would like to give this a shot if no one else is currently working on this:)
I see the java implementation has a RollingManifestWriter so I'd start there.
Feature Request / Improvement
Once we get support for merge-append https://github.com/apache/iceberg-python/pull/363 operations, and also compaction in the future, it would be good to have rolling-manifest writers that open a new file once the manifest gets to a certain size (both bytes and number of entries). This makes sure that Iceberg tables have decent parallelization when doing query planning.