cetra3 commented on code in PR #21566:
URL: https://github.com/apache/datafusion/pull/21566#discussion_r3732801374
##########
datafusion/datasource-parquet/src/opener/mod.rs:
##########
@@ -314,6 +320,109 @@ impl fmt::Debug for ParquetMorselizer {
}
}
+/// Scan-local cache for CPU-only pruning setup that can be reused across files
+/// with the same adapted expression inputs and physical schema.
+#[derive(Debug, Default)]
+pub(super) struct ParquetPruningSetupCache {
+ entries: Mutex<ParquetPruningSetupEntries>,
+}
+
+type ParquetPruningSetupEntries =
+ HashMap<ParquetPruningSetupCacheKey, ParquetPruningSetup>;
Review Comment:
This is potentially unbounded memory growth, which is not a good thing.
I think there is already precedent for caches such as `predicate_cache_size`
which I feel this should follow the same conventions.
--
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]
