refactor: bump arrow to 56

Cargo.toml

@@ -42,14 +42,14 @@ rust-version = "1.84"
 anyhow = "1.0.72"
 apache-avro = "0.17"
 array-init = "2"
-arrow-arith = { version = "55" }
-arrow-array = { version = "55" }
-arrow-buffer = { version = "55" }
-arrow-cast = { version = "55" }
-arrow-ord = { version = "55" }
-arrow-schema = { version = "55" }
-arrow-select = { version = "55" }
-arrow-string = { version = "55" }
+arrow-arith = { version = "56" }
+arrow-array = { version = "56" }
+arrow-buffer = { version = "56" }
+arrow-cast = { version = "56" }
+arrow-ord = { version = "56" }
+arrow-schema = { version = "56" }
+arrow-select = { version = "56" }
+arrow-string = { version = "56" }
 async-std = "1.12"
 async-trait = "0.1.88"
 aws-config = "1.6.1"
@@ -59,9 +59,9 @@ bytes = "1.10"
 chrono = "0.4.40"
 clap = { version = "4.5.35", features = ["derive", "cargo"] }
 ctor = "0.2.8"
-datafusion = "47"
-datafusion-cli = "47"
-datafusion-sqllogictest = "47"
+datafusion = "50"
+datafusion-cli = "50"
+datafusion-sqllogictest = "50"
 derive_builder = "0.20"
 dirs = "6"
 enum-ordinalize = "4.3.0"
@@ -76,7 +76,7 @@ iceberg = { version = "0.4.0", path = "./crates/iceberg" }
 iceberg-catalog-memory = { version = "0.4.0", path = "./crates/catalog/memory" }
 iceberg-catalog-rest = { version = "0.4.0", path = "./crates/catalog/rest" }
 iceberg-datafusion = { version = "0.4.0", path = "./crates/integrations/datafusion" }
-indicatif = "0.17"
+indicatif = "0.18"
 itertools = "0.13"
 linkedbytes = "0.1.8"
 metainfo = "0.7.14"
@@ -88,7 +88,7 @@ num-bigint = "0.4.6"
 once_cell = "1.20"
 opendal = "0.54.1"
 ordered-float = "4"
-parquet = "55"
+parquet = "56"
 pilota = "0.11.2"
 port_scanner = "0.1.5"
 pretty_assertions = "1.4"

crates/iceberg/src/arrow/reader.rs

@@ -39,7 +39,7 @@ use parquet::arrow::arrow_reader::{
 };
 use parquet::arrow::async_reader::AsyncFileReader;
 use parquet::arrow::{ParquetRecordBatchStreamBuilder, ProjectionMask, PARQUET_FIELD_ID_META_KEY};
-use parquet::file::metadata::{ParquetMetaData, ParquetMetaDataReader, RowGroupMetaData};
+use parquet::file::metadata::{PageIndexPolicy, ParquetMetaData, ParquetMetaDataReader, RowGroupMetaData};
 use parquet::schema::types::{SchemaDescriptor, Type as ParquetType};
 
 use crate::arrow::delete_file_manager::CachingDeleteFileManager;
@@ -1384,11 +1384,15 @@ impl<R: FileRead> AsyncFileReader for ArrowFileReader<R> {
         _options: Option<&'_ ArrowReaderOptions>,
     ) -> BoxFuture<'_, parquet::errors::Result<Arc<ParquetMetaData>>> {
         async move {
+            let column_policy = PageIndexPolicy::from(self.preload_column_index);
+            let offset_policy = PageIndexPolicy::from(self.preload_offset_index);
+            let page_policy = PageIndexPolicy::from(self.preload_page_index);
+
             let reader = ParquetMetaDataReader::new()
                 .with_prefetch_hint(self.metadata_size_hint)
-                .with_column_indexes(self.preload_column_index)
-                .with_page_indexes(self.preload_page_index)
-                .with_offset_indexes(self.preload_offset_index);
+                .with_column_index_policy(column_policy)
+                .with_page_index_policy(page_policy)
+                .with_offset_index_policy(offset_policy);
             let size = self.meta.size;
             let meta = reader.load_and_finish(self, size).await?;
 

crates/iceberg/src/delete_vector.rs

@@ -32,7 +32,7 @@ impl DeleteVector {
         }
     }
 
-    pub fn iter(&self) -> DeleteVectorIterator {
+    pub fn iter(&self) -> DeleteVectorIterator<'_> {
         let outer = self.inner.bitmaps();
         DeleteVectorIterator { outer, inner: None }
     }

crates/iceberg/src/inspect/metadata_table.rs

@@ -34,12 +34,12 @@ impl<'a> MetadataTable<'a> {
     }
 
     /// Get the snapshots table.
-    pub fn snapshots(&self) -> SnapshotsTable {
+    pub fn snapshots(&self) -> SnapshotsTable<'_> {
         SnapshotsTable::new(self.0)
     }
 
     /// Get the manifests table.
-    pub fn manifests(&self) -> ManifestsTable {
+    pub fn manifests(&self) -> ManifestsTable<'_> {
         ManifestsTable::new(self.0)
     }
 }

crates/integrations/datafusion/src/physical_plan/expr_to_predicate.rs

@@ -79,7 +79,7 @@ fn to_iceberg_predicate(expr: &Expr) -> TransformedResult {
             }
         }
         Expr::Column(column) => TransformedResult::Column(Reference::new(column.name())),
-        Expr::Literal(literal) => match scalar_value_to_datum(literal) {
+        Expr::Literal(literal, _file_metadata) => match scalar_value_to_datum(literal) {
             Some(data) => TransformedResult::Literal(data),
             None => TransformedResult::NotTransformed,
         },

crates/integrations/datafusion/src/physical_plan/scan.rs

@@ -102,7 +102,7 @@ impl ExecutionPlan for IcebergTableScan {
         self
     }
 
-    fn children(&self) -> Vec<&Arc<(dyn ExecutionPlan + 'static)>> {
+    fn children(&self) -> Vec<&Arc<dyn ExecutionPlan + 'static>> {
         vec![]
     }
 

rust-toolchain.toml

@@ -20,5 +20,5 @@
 #
 # The channel is exactly same day for our MSRV.
 [toolchain]
-channel = "nightly-2025-02-20"
+channel = "nightly-2025-06-27"
 components = ["rustfmt", "clippy"]