JeonDaehong commented on code in PR #3386:
URL: https://github.com/apache/iceberg-rust/pull/3386#discussion_r4236397619
##########
crates/iceberg/src/expr/visitors/page_index_evaluator.rs:
##########
@@ -548,9 +540,11 @@ impl<'a> PageIndexEvaluator<'a> {
/// Parquet stores Iceberg decimals with precision > 18 as big-endian
/// two's-complement `FIXED_LEN_BYTE_ARRAY` of width `type_length`.
///
- /// Returns an error for a bound whose width differs from `type_length`:
that
- /// signals a `column_index_truncate_length` truncation, and a truncated
byte
- /// prefix decodes to the wrong number under two's-complement.
+ /// Returns an error for a bound whose width differs from `type_length`.
+ /// Conforming writers never truncate a decimal column index (a truncated
Review Comment:
I checked this claim against both writers: `parquet` 59.3.0 (the version in
`Cargo.lock`) skips truncation for `Decimal`/`Float16` `FIXED_LEN_BYTE_ARRAY`
in `can_truncate_value`, while `parquet-java` 1.17.1's
`BinaryTruncator.getTruncator` only truncates `String`/`Enum`/`JSON`/`BSON` and
returns the no-op truncator for decimals. Therefore, the "non-conforming
writer" framing holds for files written by both `iceberg-rust` and
`iceberg-java`.
--
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]
