Enable more clippy lints

Cargo.toml

@@ -269,70 +269,61 @@ wildcard_dependencies = "warn"
 
 # Pedantic lints we opt out of, with the number of hits at the time we enabled `pedantic`.
 # Some of these we should consider enabling.
-borrow_as_ptr = "allow"                             # 6 hits
-case_sensitive_file_extension_comparisons = "allow" # 1 hit
-cast_lossless = "allow"                             # 361 hits
-cast_possible_truncation = "allow"                  # 911 hits
-cast_possible_wrap = "allow"                        # 493 hits
-cast_precision_loss = "allow"                       # 266 hits
-cast_ptr_alignment = "allow"                        # 5 hits
-cast_sign_loss = "allow"                            # 440 hits
-cloned_instead_of_copied = "allow"                  # 38 hits
-default_trait_access = "allow"                      # 221 hits
-doc_comment_double_space_linebreaks = "allow"       # 6 hits
-doc_link_with_quotes = "allow"                      # 25 hits
-doc_markdown = "allow"                              # 4933 hits; needs a long `doc-valid-idents` list in `clippy.toml`
-enum_glob_use = "allow"                             # 98 hits
-explicit_into_iter_loop = "allow"                   # 55 hits
-explicit_iter_loop = "allow"                        # 189 hits
-float_cmp = "allow"                                 # 8 hits; exact float comparisons are often intentional here
-format_collect = "allow"                            # 4 hits
-format_push_string = "allow"                        # 34 hits
-from_iter_instead_of_collect = "allow"              # 51 hits
-if_not_else = "allow"                               # 133 hits
-ignored_unit_patterns = "allow"                     # 52 hits
-implicit_clone = "allow"                            # 198 hits
-implicit_hasher = "allow"                           # 17 hits
-inline_always = "allow"                             # 45 hits
-items_after_statements = "allow"                    # 171 hits
-large_digit_groups = "allow"                        # 3 hits
-manual_string_new = "allow"                         # 84 hits
-many_single_char_names = "allow"                    # 12 hits; short names are idiomatic in the numeric kernels
-map_unwrap_or = "allow"                             # 198 hits
-match_bool = "allow"                                # 46 hits
-match_same_arms = "allow"                           # 261 hits
-match_wildcard_for_single_variants = "allow"        # 132 hits
-missing_errors_doc = "allow"                        # 1807 hits
-missing_fields_in_debug = "allow"                   # 29 hits
-missing_panics_doc = "allow"                        # 244 hits
-must_use_candidate = "allow"                        # 2726 hits
-needless_bitwise_bool = "allow"                     # 1 hit
-needless_continue = "allow"                         # 37 hits
-needless_raw_string_hashes = "allow"                # 540 hits
-no_effect_underscore_binding = "allow"              # 1 hit
-ptr_as_ptr = "allow"                                # 83 hits
-redundant_closure_for_method_calls = "allow"        # 686 hits
-redundant_else = "allow"                            # 48 hits
-ref_option = "allow"                                # 36 hits
-return_self_not_must_use = "allow"                  # 644 hits
-semicolon_if_nothing_returned = "allow"             # 1353 hits
-similar_names = "allow"                             # 228 hits; too many false positives, e.g. `expr`/`exprs`
-single_char_pattern = "allow"                       # 23 hits
-single_match_else = "allow"                         # 155 hits
-struct_excessive_bools = "allow"                    # 24 hits
-struct_field_names = "allow"                        # 14 hits
-too_many_lines = "allow"                            # 484 hits
-trivially_copy_pass_by_ref = "allow"                # 74 hits
-unicode_not_nfc = "allow"                           # 2 hits
-unnecessary_literal_bound = "allow"                 # 471 hits
-unnecessary_semicolon = "allow"                     # 185 hits
-unnecessary_trailing_comma = "allow"                # 49 hits
-unnecessary_wraps = "allow"                         # 427 hits
-unnested_or_patterns = "allow"                      # 68 hits
-unreadable_literal = "allow"                        # 502 hits
-unused_self = "allow"                               # 69 hits
-used_underscore_items = "allow"                     # 28 hits
-wildcard_imports = "allow"                          # 48 hits; `use crate::prelude::*` is idiomatic
+borrow_as_ptr = "allow"                       # 6 hits
+cast_lossless = "allow"                       # 361 hits
+cast_possible_truncation = "allow"            # 911 hits
+cast_possible_wrap = "allow"                  # 493 hits
+cast_precision_loss = "allow"                 # 266 hits
+cast_sign_loss = "allow"                      # 440 hits
+default_trait_access = "allow"                # 221 hits
+doc_comment_double_space_linebreaks = "allow" # 6 hits
+doc_markdown = "allow"                        # 4933 hits; needs a long `doc-valid-idents` list in `clippy.toml`
+enum_glob_use = "allow"                       # 98 hits
+explicit_into_iter_loop = "allow"             # 55 hits
+explicit_iter_loop = "allow"                  # 189 hits
+float_cmp = "allow"                           # 8 hits; exact float comparisons are often intentional here
+format_collect = "allow"                      # 4 hits
+format_push_string = "allow"                  # 34 hits
+from_iter_instead_of_collect = "allow"        # 51 hits
+if_not_else = "allow"                         # 133 hits
+ignored_unit_patterns = "allow"               # 52 hits
+implicit_clone = "allow"                      # 198 hits
+implicit_hasher = "allow"                     # 17 hits
+inline_always = "allow"                       # 45 hits
+items_after_statements = "allow"              # 171 hits
+large_digit_groups = "allow"                  # 3 hits
+manual_string_new = "allow"                   # 84 hits
+many_single_char_names = "allow"              # 12 hits; short names are idiomatic in the numeric kernels
+map_unwrap_or = "allow"                       # 198 hits
+match_bool = "allow"                          # 46 hits
+match_same_arms = "allow"                     # 261 hits
+match_wildcard_for_single_variants = "allow"  # 132 hits
+missing_errors_doc = "allow"                  # 1807 hits
+missing_fields_in_debug = "allow"             # 29 hits
+missing_panics_doc = "allow"                  # 244 hits
+must_use_candidate = "allow"                  # 2726 hits
+needless_bitwise_bool = "allow"               # 1 hit
+needless_raw_string_hashes = "allow"          # 540 hits
+ptr_as_ptr = "allow"                          # 83 hits
+redundant_closure_for_method_calls = "allow"  # 686 hits
+redundant_else = "allow"                      # 48 hits
+return_self_not_must_use = "allow"            # 644 hits
+semicolon_if_nothing_returned = "allow"       # 1353 hits
+similar_names = "allow"                       # 228 hits; too many false positives, e.g. `expr`/`exprs`
+single_char_pattern = "allow"                 # 23 hits
+single_match_else = "allow"                   # 155 hits
+struct_excessive_bools = "allow"              # 24 hits
+struct_field_names = "allow"                  # 14 hits
+too_many_lines = "allow"                      # 484 hits
+trivially_copy_pass_by_ref = "allow"          # 74 hits
+unicode_not_nfc = "allow"                     # 2 hits
+unnecessary_literal_bound = "allow"           # 471 hits
+unnecessary_wraps = "allow"                   # 427 hits
+unnested_or_patterns = "allow"                # 68 hits
+unreadable_literal = "allow"                  # 502 hits
+unused_self = "allow"                         # 69 hits
+used_underscore_items = "allow"               # 28 hits
+wildcard_imports = "allow"                    # 48 hits; `use crate::prelude::*` is idiomatic
 
 [workspace.lints.rust]
 unexpected_cfgs = { level = "warn", check-cfg = [

benchmarks/src/imdb/mod.rs

@@ -233,7 +233,7 @@ pub fn get_query_sql(query: &str) -> Result<Vec<String>> {
                     .collect());
             }
             Err(e) => errors.push(format!("{filename}: {e}")),
-        };
+        }
     }
     plan_err!("invalid query. Could not find query: {:?}", errors)
 }

benchmarks/src/tpcds/run.rs

@@ -142,7 +142,7 @@ pub fn get_query_sql(base_query_path: &str, query: usize) -> Result<Vec<String>>
                     .collect());
             }
             Err(e) => errors.push(format!("{filename}: {e}")),
-        };
+        }
 
         plan_err!("invalid query. Could not find query: {:?}", errors)
     } else {

benchmarks/src/tpch/mod.rs

@@ -209,7 +209,7 @@ pub fn get_query_sql_for_scale_factor(
                         .collect());
                 }
                 Err(e) => errors.push(format!("{filename}: {e}")),
-            };
+            }
         }
         plan_err!("invalid query. Could not find query: {:?}", errors)
     } else {

benchmarks/src/util/run.rs

@@ -231,7 +231,7 @@ impl BenchmarkRun {
     pub fn maybe_write_json(&self, maybe_path: Option<impl AsRef<Path>>) -> Result<()> {
         if let Some(path) = maybe_path {
             std::fs::write(path, self.to_json())?;
-        };
+        }
         Ok(())
     }
 }

datafusion-cli/src/catalog.rs

@@ -180,7 +180,7 @@ impl SchemaProvider for DynamicObjectStoreSchemaProvider {
                         }
                     }
                     _ => {}
-                };
+                }
                 state = builder.build();
                 let store = get_object_store(
                     &state,

datafusion-cli/src/exec.rs

@@ -73,12 +73,7 @@ pub async fn exec_from_lines(
 
     for line in reader.lines() {
         match line {
-            Ok(line) if line.starts_with("#!") => {
-                continue;
-            }
-            Ok(line) if line.starts_with("--") => {
-                continue;
-            }
+            Ok(line) if line.starts_with("#!") || line.starts_with("--") => {}
             Ok(line) => {
                 let line = line.trim_end();
                 query.push_str(line);
@@ -197,7 +192,6 @@ pub async fn exec_from_repl(
             Err(ReadlineError::Interrupted) => {
                 println!("^C");
                 rl.helper().unwrap().reset_hint();
-                continue;
             }
             Err(ReadlineError::Eof) => {
                 println!("\\q");

datafusion-cli/src/functions.rs

@@ -416,7 +416,7 @@ impl TableFunctionImpl for ParquetMetadataFunc {
                     stats_distinct_count_arr.push(None);
                     stats_min_value_arr.push(None);
                     stats_max_value_arr.push(None);
-                };
+                }
                 compression_arr.push(format!("{:?}", column.compression()));
                 // need to collect into Vec to format
                 let encodings: Vec<_> = column.encodings().collect();

datafusion-cli/src/main.rs

@@ -208,7 +208,7 @@ async fn main_inner() -> Result<()> {
     if let Some(ref path) = args.data_path {
         let p = Path::new(path);
         env::set_current_dir(p).unwrap();
-    };
+    }
 
     let session_config = get_session_config(&args)?;
 
@@ -353,7 +353,7 @@ fn get_session_config(args: &Args) -> Result<SessionConfig> {
         }
         config_options.execution.batch_size =
             datafusion_common::config::ConfigNonZeroUsize::try_new(batch_size)?;
-    };
+    }
 
     // use easier to understand "tree" mode by default
     // if the user hasn't specified an explain format in the environment

datafusion-cli/src/print_options.rs

@@ -151,7 +151,7 @@ impl PrintOptions {
             return Err(DataFusionError::External(
                 "PrintFormat::Table is not implemented".to_string().into(),
             ));
-        };
+        }
 
         let stdout = io::stdout();
         let mut writer = stdout.lock();

datafusion-examples/examples/data_io/parquet_embedded_index.rs

@@ -367,7 +367,7 @@ fn read_distinct_index(path: &Path) -> Result<DistinctIndex> {
     let file = File::open(path)?;
 
     let file_size = file.metadata()?.len();
-    println!("Reading index from {} (size: {file_size})", path.display(),);
+    println!("Reading index from {} (size: {file_size})", path.display());
 
     let reader = SerializedFileReader::new(file.try_clone()?)?;
     let meta = reader.metadata().file_metadata();

datafusion-examples/examples/data_io/parquet_index.rs

@@ -311,7 +311,7 @@ impl Display for ParquetMetadataIndex {
             self.last_num_pruned()
         )?;
         let batches = pretty_format_batches(std::slice::from_ref(&self.index)).unwrap();
-        write!(f, "{batches}",)
+        write!(f, "{batches}")
     }
 }
 

datafusion-examples/examples/query_planning/pruning.rs

@@ -211,5 +211,5 @@ fn create_pruning_predicate(expr: Expr, schema: &SchemaRef) -> PruningPredicate
 }
 
 fn i32_array<'a>(values: impl Iterator<Item = &'a Option<i32>>) -> ArrayRef {
-    Arc::new(Int32Array::from_iter(values.cloned()))
+    Arc::new(Int32Array::from_iter(values.copied()))
 }

datafusion-examples/examples/query_planning/thread_pools.rs

@@ -305,7 +305,7 @@ impl Drop for CpuRuntime {
             // If the thread is still running, we wait for it to finish
             print!("Shutting down CPU runtime thread...");
             if let Err(e) = thread_join_handle.join() {
-                eprintln!("Error joining CPU runtime thread: {e:?}",);
+                eprintln!("Error joining CPU runtime thread: {e:?}");
             } else {
                 println!("CPU runtime thread shutdown successfully.");
             }

datafusion/catalog-listing/src/helpers.rs

@@ -413,7 +413,7 @@ pub async fn pruned_partition_list<'a>(
             .try_filter_map(|object_meta| {
                 futures::future::ready(object_meta_to_partitioned_file(
                     object_meta,
-                    table_path.get_table_ref(),
+                    table_path.get_table_ref().as_ref(),
                 ))
             })
             .boxed())
@@ -443,7 +443,7 @@ pub async fn pruned_partition_list<'a>(
 
 fn object_meta_to_partitioned_file(
     object_meta: ObjectMeta,
-    table_ref: &Option<TableReference>,
+    table_ref: Option<&TableReference>,
 ) -> Result<Option<PartitionedFile>> {
     Ok(Some(PartitionedFile {
         object_meta,
@@ -454,7 +454,7 @@ fn object_meta_to_partitioned_file(
         ordering: None,
         extensions: FileExtensions::new(),
         metadata_size_hint: None,
-        table_reference: table_ref.clone(),
+        table_reference: table_ref.cloned(),
     }))
 }
 

datafusion/catalog-listing/src/table.rs

@@ -673,7 +673,7 @@ impl ListingTable {
                 }
             }
             None => {} // no ordering required
-        };
+        }
 
         let output_partitioning = if let Some(output_partitioning) =
             declared_output_partitioning

datafusion/catalog/src/dynamic_file/catalog.rs

@@ -138,7 +138,7 @@ impl SchemaProvider for DynamicFileSchemaProvider {
     ) -> datafusion_common::Result<Option<Arc<dyn TableProvider>>> {
         if let Some(table) = self.inner.table(name).await? {
             return Ok(Some(table));
-        };
+        }
 
         self.factory.try_new(name).await
     }

datafusion/common/src/column.rs

@@ -236,7 +236,7 @@ impl Column {
                 .flat_map(|s| s.qualified_fields_with_unqualified_name(&self.name))
                 .collect::<Vec<_>>();
             match qualified_fields.len() {
-                0 => continue,
+                0 => {}
                 1 => return Ok(Column::from(qualified_fields[0])),
                 _ => {
                     // More than 1 fields in this schema have their names set to self.name.

datafusion/common/src/config.rs

@@ -3283,7 +3283,7 @@ impl ConfigField for ConfigFileEncryptionProperties {
         if key.contains("::") {
             // Handle any column specific properties
             return self.column_encryption_properties.set(key, value);
-        };
+        }
 
         let (key, rem) = key.split_once('.').unwrap_or((key, ""));
         match key {
@@ -3463,7 +3463,7 @@ impl ConfigField for ConfigFileDecryptionProperties {
         if key.contains("::") {
             // Handle any column specific properties
             return self.column_decryption_properties.set(key, value);
-        };
+        }
 
         let (key, rem) = key.split_once('.').unwrap_or((key, ""));
         match key {

datafusion/common/src/error.rs

@@ -1311,7 +1311,7 @@ mod test {
         match std::env::var("RUST_BACKTRACE") {
             Ok(val) if val == "1" => {}
             _ => panic!("Environment variable RUST_BACKTRACE must be set to 1"),
-        };
+        }
     }
 
     // To pass the test the environment variable RUST_BACKTRACE should be set to 1 to enforce backtrace