kotwal-itpro commented on PR #8802: URL: https://github.com/apache/hop/pull/8802#issuecomment-6067654714
Thanks @bamaer, all four were real. Fixed in fe6ca55400: **S3 and MinIO.** `HopVfsSeekableInputStream` now catches the `FileSystemException` from `getRandomAccessContent` and falls back to reopening and skipping. It only tries random access once per stream, so a provider like S3 doesn't throw on every seek. `HopVfsSeekableInputStreamTest` uses a file system that reports `RANDOM_ACCESS_READ` and throws `random-access-not-supported`, like `S3CommonFileObject`; it fails without the fix. **Dates and timestamps without a zone.** Read in the JVM zone like `ParquetValueConverter`: `atStartOfDay(ZoneId.systemDefault())` for dates and `Timestamp.valueOf(LocalDateTime)` for timestamps. `toIcebergValue` does the reverse, so a value written and read back is unchanged. `IcebergTypeMapperTest` runs the round trips in UTC, Los Angeles, Berlin, Kolkata and Kiritimati (UTC+14), and `LakeTableInputLocalTest` now reads a table in three zones. Without the fix the date comes back as 2026-09-30 17:00 in Los Angeles, as you said. Those tests also found one more bug: writing a `timestamp_ns` value failed with a `ClassCastException`, because `TimestampNanoType` isn't a `TimestampType`. Fixed as well. **ORC and Avro data files** now fail with "Data file ... is in AVRO format. Only Parquet data files are supported on the local engine; ORC and Avro tables can be read with the Spark engine." There's a test with a real Avro data file. **Snapshot across copies.** With more than one copy, the first copy to start resolves the current snapshot and stores it in the pipeline's extension data map; the other copies use that id. `copiesShareTheCurrentSnapshot` commits between the first and the second copy resolving it, and both get the same snapshot. Time travel was already fixed per run, so this only applies to reading the current snapshot. Tests: lakehouse 60 (was 41), Spark 179, RAT clean. I added the fixes as a separate commit to keep them easy to review; happy to squash before merge. -- 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]
