laskoviymishka opened a new pull request, #1270: URL: https://github.com/apache/iceberg-go/pull/1270
matchDVToData decided DV applicability with `dataEntry.SequenceNum() <= dvEntry.SequenceNum()`. A deletion-vector ManifestEntry can reach this point with an unset sequence number, for which SequenceNum() reports the -1 sentinel — so for any real (>= 0) data sequence number the comparison dataSeq <= -1 was always false and the DV was silently skipped, resurfacing rows it should have deleted. Entries arrive here already inherited, so a committed ADDED entry always carries a real sequence number; an unset value comes from an EXISTING/DELETED entry missing its required explicit sequence number, or a not-yet-committed manifest. Treat an indeterminate (negative) sequence number as applicable — mirroring the existing seq >= 0 sentinel guard in the scan task builder — so an unknown sequence never drops a DV. -- 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]
