mhilton commented on code in PR #7738:
URL: https://github.com/apache/arrow-datafusion/pull/7738#discussion_r1345851891
##########
datafusion/core/src/physical_optimizer/pruning.rs:
##########
@@ -146,17 +172,14 @@ impl PruningPredicate {
///
/// `true`: There MAY be rows that match the predicate
///
- /// `false`: There are no rows that could match the predicate
+ /// `false`: There are no rows that could possible match the predicate
Review Comment:
*possibly
##########
datafusion/core/src/physical_optimizer/pruning.rs:
##########
@@ -91,10 +83,44 @@ pub trait PruningStatistics {
fn null_counts(&self, column: &Column) -> Option<ArrayRef>;
}
-/// Evaluates filter expressions on statistics in order to
-/// prune data containers (e.g. parquet row group)
+/// Evaluates filter expressions on statistics, rather than the actual data. If
+/// no rows could possibly pass the filter entire containers can be "pruned"
+/// (skipped), without reading any actual data, leading to significant
+/// performance improvements.
+///
+/// [`PruningPredicate`]s are used to prune (avoid scanning) Parquet Row Groups
+/// based on the min/max values found in the Parquet metadata. If the
+/// `PruningPredicate` can guarantee that no rows in the Row Group match the
+/// filter, the entire Row Group is skipped during query execution.
+///
+/// Note that this API is designed to be general, as it works:
+///
+/// 1. Arbitrary expressions expressions (including user defined functions)
+///
+/// 2. Anything that implements the [`PruningStatistics`] interface, not just
Review Comment:
Should "interface" be "trait" in this context?
--
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]
