zeroshade commented on PR #1013: URL: https://github.com/apache/arrow-go/pull/1013#issuecomment-5094392637
Reviewed and the analysis holds up: `DeltaLengthByteArrayDecoder.Decode` indexes `d.lengths[i]` with `nvals` derived from the *logical* count, so a nullable page with two logical positions and one physical value reads past the end of the length table. Switching to the encoded physical count plus the negative/oversized length validation is the right fix. I just merged #1014, which rewrites the same `DeltaByteArrayDecoder.SetData` this PR touches, so this now needs a rebase onto `main` — the hunks overlap directly. Once you rebase and CI is green I will merge this. Note that #1014 already added the `lastVal` reset and the prefix-offset bounds check, so some of the overlap should collapse. -- 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]
