Zach Dischner
Zach Dischner
## Rev 2 * Default implementation in `RewriteManifests` interface * Switched to using hidden transform names to specify sort order, not raw column name * Addressed other PR feedback Going...
> Looks like the CI test failed, @zachdisc could you take a look and fix? > > Overall, we have the 2 options based on this thread: [#9731 (comment)](https://github.com/apache/iceberg/pull/9731#discussion_r1494774386) >...
## Rev 3 * Fixed styling which caused CI/CD to fail * Changed the clustering column to be a struct referencing the actual columns instead of a contacted string `p1::p2::p3`....
## Rev 3 * Addressed last round of comments * Added a new API option to sort with a custom supplied `Function`. Illustration (in unit test) - say you have...
@nastra nastra > @zachdisc Hey Zach, sorry I was trying to experiment with refactoring the current PR, but somehow the branch is public and I ended up directly committing to...
> I see that's a good point, using the UDF function definitely is less performant than sorting native columns, in that case we can keep it as is, unless anyone...
#### R4 Renamed `sort` to `clusterBy` - more honest and aligns with the core API as well. Have not made progress on using the custom UDF for both standard partition...
@jackye1995 @nastra any comments or thoughts on this?
Made most of the changes asked for. Still undecided around clustering vs sorting naming, and I didn't change the test that verifies content ordering within manifests.
@RussellSpitzer mind a review?