mxm opened a new pull request, #17189: URL: https://github.com/apache/iceberg/pull/17189
`ConvertEqualityDeletes` has two operational modes: 1. "in-place", where equality deletes are written directly to the target branch and subsequently converted to deletion vectors on the same branch. 2. Separate staging branch, where a staging branch holds equality deletes before they are converted and committed to a main/target branch. When `ConvertEqualityDeletes` is run in "in-place" mode, equality deletes are written directly on the target branch. We then convert them to deletion vectors, but leave the equality delete files in the resulting snapshot. Consequently, readers will still load and apply them and perform merge-on-read, which is redundant at this point because the DVs written already cover the rows the equality deletes would remove. Hence, this change removes the equality delete files in the same commit that writes the DVs. Separate branch mode is unchanged: after converting equality deletes from the staging branch, we commit DVs to a target branch which remains free of equality deletes. -- 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]
