tanmayrauth commented on PR #891:
URL: https://github.com/apache/iceberg-go/pull/891#issuecomment-4247413621
1. Float normalization — Added -0.0 → 0.0 and NaN → canonical NaN
normalization before bit-encoding both float32 and float64. Without this, -0.0
and 0.0 produce different bit patterns and different hashes, causing
false-negative bloom filter results (row group skipped when value IS present).
2. FixedLiteral type safety— Replaced the direct type assertion
lit.(iceberg.TypedLiteral[[]byte]) with a type switch that handles both
TypedLiteral[[]byte] and iceberg.FixedLiteral directly, avoiding a potential
runtime panic.
3. Bloom-only pruning— Changed the guard from StatsFn != nil to StatsFn !=
nil || len(BloomPreds) > 0, so bloom-only pruning works when StatsFn is nil.
Inside the loop, StatsFn is now called conditionally (use := true default), so
a nil stats function is treated as "keep all row groups."
--
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]
