alexey-bass commented on PR #13276: URL: https://github.com/apache/seatunnel/pull/13276#issuecomment-7023597872
Thanks for the review. Updated in `9a501b4`: - Issue 1: the comment on `MAX_JOB_NAME_BYTES` now says the limit only decides which decoder runs and does not cap the output. Truncating the name changes what the endpoint returns, so I would leave it to a follow-up. - Issue 2: added the three cases. Length -2 and length 1000 in data that ends 4 bytes later are rejected by the prefix decoder. A valid `Data` with the versioned flag and two version bytes decodes (`shouldSkipVersionBytesInFastPath`); Hazelcast's own deserialization reads the same bytes in that test. Removing either length guard or the versioned branch now fails a test. - Issue 4: `@param` / `@throws` added, and a comment in `JobImmutableInformation#writeData` points to the prefix reader. - Issue 5: every rejection logs at debug with the job id, the failed check and `data.totalSize()`. - Issue 6: the legacy test calls `decodeJobBasicInfoFromPrefix`. - Issue 7: padding is 9 MiB, with a comment that it is serialized twice. - Issue 8: `NodeExtension` is imported. Issue 3 is not in this commit: it predates the PR and changes what the cache holds, so I would do it as a follow-up. `JobInfoServiceNullSafetyTest` passes locally (11 tests), spotless check passes. The fork Build for the new head is running: https://github.com/alexey-bass/seatunnel/actions/runs/44625704001 -- 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]
