Upgrade Guides¶

DataFusion `49.0.0`¶

Note: DataFusion 49.0.0 has not been released yet. The information provided in this section pertains to features and changes that have already been merged to the main branch and are awaiting release in this version. You can see the current status of the 49.0.0 release here

`datafusion.execution.collect_statistics` now defaults to `true`¶

The default value of the datafusion.execution.collect_statistics configuration setting is now true. This change impacts users that use that value directly and relied on its default value being false.

This change also restores the default behavior of ListingTable to its previous. If you use it directly you can maintain the current behavior by overriding the default value in your code.

ListingOptions::new(Arc::new(ParquetFormat::default()))
    .with_collect_stat(false)
    // other options

Metadata is now represented by `FieldMetadata`¶

Metadata from the Arrow Field is now stored using the FieldMetadata structure. In prior versions it was stored as both a HashMap<String, String> and a BTreeMap<String, String>. FieldMetadata is a easier to work with and is more efficient.

To create FieldMetadata from a Field:

 let metadata = FieldMetadata::from(&field);

To add metadata to a Field, use the add_to_field method:

let updated_field = metadata.add_to_field(field);

See #16317 for details.

New `datafusion.execution.spill_compression` configuration option¶

DataFusion 49.0.0 adds support for compressing spill files when data is written to disk during spilling query execution. A new configuration option datafusion.execution.spill_compression controls the compression codec used.

Configuration:

Key: datafusion.execution.spill_compression
Default: uncompressed
Valid values: uncompressed, lz4_frame, zstd

Usage:

use datafusion::prelude::*;
use datafusion_common::config::SpillCompression;

let config = SessionConfig::default()
    .with_spill_compression(SpillCompression::Zstd);
let ctx = SessionContext::new_with_config(config);

Or via SQL:

SET datafusion.execution.spill_compression = 'zstd';

For more details about this configuration option, including performance trade-offs between different compression codecs, see the Configuration Settings documentation.

DataFusion `48.0.0`¶

`Expr::Literal` has optional metadata¶

The Expr::Literal variant now includes optional metadata, which allows for carrying through Arrow field metadata to support extension types and other uses.

This means code such as

match expr {
...
  Expr::Literal(scalar) => ...
...
}

Should be updated to:

match expr {
...
  Expr::Literal(scalar, _metadata) => ...
...
}

Likewise constructing Expr::Literal requires metadata as well. The lit function has not changed and returns an Expr::Literal with no metadata.

`Expr::WindowFunction` is now `Box`ed¶

Expr::WindowFunction is now a Box<WindowFunction> instead of a WindowFunction directly. This change was made to reduce the size of Expr and improve performance when planning queries (see details on #16207).

This is a breaking change, so you will need to update your code if you match on Expr::WindowFunction directly. For example, if you have code like this:

match expr {
  Expr::WindowFunction(WindowFunction {
    params:
      WindowFunctionParams {
       partition_by,
       order_by,
      ..
    }
  }) => {
    // Use partition_by and order_by as needed
  }
  _ => {
    // other expr
  }
}

You will need to change it to:

match expr {
  Expr::WindowFunction(window_fun) => {
    let WindowFunction {
      fun,
      params: WindowFunctionParams {
        args,
        partition_by,
        ..
        },
    } = window_fun.as_ref();
    // Use partition_by and order_by as needed
  }
  _ => {
    // other expr
  }
}

The `VARCHAR` SQL type is now represented as `Utf8View` in Arrow¶

The mapping of the SQL VARCHAR type has been changed from Utf8 to Utf8View which improves performance for many string operations. You can read more about Utf8View in the DataFusion blog post on German-style strings

This means that when you create a table with a VARCHAR column, it will now use Utf8View as the underlying data type. For example:

> CREATE TABLE my_table (my_column VARCHAR);
0 row(s) fetched.
Elapsed 0.001 seconds.

> DESCRIBE my_table;
+-------------+-----------+-------------+
| column_name | data_type | is_nullable |
+-------------+-----------+-------------+
| my_column   | Utf8View  | YES         |
+-------------+-----------+-------------+
1 row(s) fetched.
Elapsed 0.000 seconds.

You can restore the old behavior of using Utf8 by changing the datafusion.sql_parser.map_varchar_to_utf8view configuration setting. For example

> set datafusion.sql_parser.map_varchar_to_utf8view = false;
0 row(s) fetched.
Elapsed 0.001 seconds.

> CREATE TABLE my_table (my_column VARCHAR);
0 row(s) fetched.
Elapsed 0.014 seconds.

> DESCRIBE my_table;
+-------------+-----------+-------------+
| column_name | data_type | is_nullable |
+-------------+-----------+-------------+
| my_column   | Utf8      | YES         |
+-------------+-----------+-------------+
1 row(s) fetched.
Elapsed 0.004 seconds.

`ListingOptions` default for `collect_stat` changed from `true` to `false`¶

This makes it agree with the default for SessionConfig. Most users won’t be impacted by this change but if you were using ListingOptions directly and relied on the default value of collect_stat being true, you will need to explicitly set it to true in your code.

ListingOptions::new(Arc::new(ParquetFormat::default()))
    .with_collect_stat(true)
    // other options

Processing `FieldRef` instead of `DataType` for user defined functions¶

In order to support metadata handling and extension types, user defined functions are now switching to traits which use FieldRef rather than a DataType and nullability. This gives a single interface to both of these parameters and additionally allows access to metadata fields, which can be used for extension types.

To upgrade structs which implement ScalarUDFImpl, if you have implemented return_type_from_args you need instead to implement return_field_from_args. If your functions do not need to handle metadata, this should be straightforward repackaging of the output data into a FieldRef. The name you specify on the field is not important. It will be overwritten during planning. ReturnInfo has been removed, so you will need to remove all references to it.

ScalarFunctionArgs now contains a field called arg_fields. You can use this to access the metadata associated with the columnar values during invocation.

To upgrade user defined aggregate functions, there is now a function return_field that will allow you to specify both metadata and nullability of your function. You are not required to implement this if you do not need to handle metatdata.

The largest change to aggregate functions happens in the accumulator arguments. Both the AccumulatorArgs and StateFieldsArgs now contain FieldRef rather than DataType.

To upgrade window functions, ExpressionArgs now contains input fields instead of input data types. When setting these fields, the name of the field is not important since this gets overwritten during the planning stage. All you should need to do is wrap your existing data types in fields with nullability set depending on your use case.

Physical Expression return `Field`¶

To support the changes to user defined functions processing metadata, the PhysicalExpr trait, which now must specify a return Field based on the input schema. To upgrade structs which implement PhysicalExpr you need to implement the return_field function. There are numerous examples in the physical-expr crate.

`FileFormat::supports_filters_pushdown` replaced with `FileSource::try_pushdown_filters`¶

To support more general filter pushdown, the FileFormat::supports_filters_pushdown was replaced with FileSource::try_pushdown_filters. If you implemented a custom FileFormat that uses a custom FileSource you will need to implement FileSource::try_pushdown_filters. See ParquetSource::try_pushdown_filters for an example of how to implement this.

FileFormat::supports_filters_pushdown has been removed.

`ParquetExec`, `AvroExec`, `CsvExec`, `JsonExec` Removed¶

ParquetExec, AvroExec, CsvExec, and JsonExec were deprecated in DataFusion 46 and are removed in DataFusion 48. This is sooner than the normal process described in the API Deprecation Guidelines because all the tests cover the new DataSourceExec rather than the older structures. As we evolve DataSource, the old structures began to show signs of “bit rotting” (not working but no one knows due to lack of test coverage).

`PartitionedFile` added as an argument to the `FileOpener` trait¶

This is necessary to properly fix filter pushdown for filters that combine partition columns and file columns (e.g. day = username['dob']).

If you implemented a custom FileOpener you will need to add the PartitionedFile argument but are not required to use it in any way.

DataFusion `47.0.0`¶

This section calls out some of the major changes in the 47.0.0 release of DataFusion.

Here are some example upgrade PRs that demonstrate changes required when upgrading from DataFusion 46.0.0:

Upgrades to `arrow-rs` and `arrow-parquet` 55.0.0 and `object_store` 0.12.0¶

Several APIs are changed in the underlying arrow and parquet libraries to use a u64 instead of usize to better support WASM (See #7371 and [#6961])

Additionally ObjectStore::list and ObjectStore::list_with_offset have been changed to return static lifetimes (See #6619)

This requires converting from usize to u64 occasionally as well as changes to ObjectStore implementations such as

impl Objectstore {
    ...
    // The range is now a u64 instead of usize
    async fn get_range(&self, location: &Path, range: Range<u64>) -> ObjectStoreResult<Bytes> {
        self.inner.get_range(location, range).await
    }
    ...
    // the lifetime is now 'static instead of `_ (meaning the captured closure can't contain references)
    // (this also applies to list_with_offset)
    fn list(&self, prefix: Option<&Path>) -> BoxStream<'static, ObjectStoreResult<ObjectMeta>> {
        self.inner.list(prefix)
    }
}

The ParquetObjectReader has been updated to no longer require the object size (it can be fetched using a single suffix request). See #7334 for details

Pattern in DataFusion 46.0.0:

let meta: ObjectMeta = ...;
let reader = ParquetObjectReader::new(store, meta);

Pattern in DataFusion 47.0.0:

let meta: ObjectMeta = ...;
let reader = ParquetObjectReader::new(store, location)
  .with_file_size(meta.size);

`DisplayFormatType::TreeRender`¶

DataFusion now supports tree style explain plans. Implementations of Executionplan must also provide a description in the DisplayFormatType::TreeRender format. This can be the same as the existing DisplayFormatType::Default.

Removed Deprecated APIs¶

Several APIs have been removed in this release. These were either deprecated previously or were hard to use correctly such as the multiple different ScalarUDFImpl::invoke* APIs. See #15130, #15123, and #15027 for more details.

`FileScanConfig` –> `FileScanConfigBuilder`¶

Previously, FileScanConfig::build() directly created ExecutionPlans. In DataFusion 47.0.0 this has been changed to use FileScanConfigBuilder. See #15352 for details.

Pattern in DataFusion 46.0.0:

let plan = FileScanConfig::new(url, schema, Arc::new(file_source))
  .with_statistics(stats)
  ...
  .build()

Pattern in DataFusion 47.0.0:

let config = FileScanConfigBuilder::new(url, schema, Arc::new(file_source))
  .with_statistics(stats)
  ...
  .build();
let scan = DataSourceExec::from_data_source(config);

DataFusion `46.0.0`¶

Use `invoke_with_args` instead of `invoke()` and `invoke_batch()`¶

DataFusion is moving to a consistent API for invoking ScalarUDFs, ScalarUDFImpl::invoke_with_args(), and deprecating ScalarUDFImpl::invoke(), ScalarUDFImpl::invoke_batch(), and ScalarUDFImpl::invoke_no_args()

If you see errors such as the following it means the older APIs are being used:

This feature is not implemented: Function concat does not implement invoke but called

To fix this error, use ScalarUDFImpl::invoke_with_args() instead, as shown below. See PR 14876 for an example.

Given existing code like this:

impl ScalarUDFImpl for SparkConcat {
...
    fn invoke_batch(&self, args: &[ColumnarValue], number_rows: usize) -> Result<ColumnarValue> {
        if args
            .iter()
            .any(|arg| matches!(arg.data_type(), DataType::List(_)))
        {
            ArrayConcat::new().invoke_batch(args, number_rows)
        } else {
            ConcatFunc::new().invoke_batch(args, number_rows)
        }
    }
}

impl ScalarUDFImpl for SparkConcat {
    ...
    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
        if args
            .args
            .iter()
            .any(|arg| matches!(arg.data_type(), DataType::List(_)))
        {
            ArrayConcat::new().invoke_with_args(args)
        } else {
            ConcatFunc::new().invoke_with_args(args)
        }
    }
}

`ParquetExec`, `AvroExec`, `CsvExec`, `JsonExec` deprecated¶

DataFusion 46 has a major change to how the built in DataSources are organized. Instead of individual ExecutionPlans for the different file formats they now all use DataSourceExec and the format specific information is embodied in new traits DataSource and FileSource.

Here is more information about

Design Ticket
Change PR PR #14224
Example of an Upgrade PR in delta-rs

Cookbook: Changes to `ParquetExecBuilder`¶

Code that looks for ParquetExec like this will no longer work:

    if let Some(parquet_exec) = plan.as_any().downcast_ref::<ParquetExec>() {
        // Do something with ParquetExec here
    }

Instead, with DataSourceExec, the same information is now on FileScanConfig and ParquetSource. The equivalent code is

if let Some(datasource_exec) = plan.as_any().downcast_ref::<DataSourceExec>() {
  if let Some(scan_config) = datasource_exec.data_source().as_any().downcast_ref::<FileScanConfig>() {
    // FileGroups, and other information is on the FileScanConfig
    // parquet
    if let Some(parquet_source) = scan_config.file_source.as_any().downcast_ref::<ParquetSource>()
    {
      // Information on PruningPredicates and parquet options are here
    }
}

Cookbook: Changes to `ParquetExecBuilder`¶

Likewise code that builds ParquetExec using the ParquetExecBuilder such as the following must be changed:

let mut exec_plan_builder = ParquetExecBuilder::new(
    FileScanConfig::new(self.log_store.object_store_url(), file_schema)
        .with_projection(self.projection.cloned())
        .with_limit(self.limit)
        .with_table_partition_cols(table_partition_cols),
)
.with_schema_adapter_factory(Arc::new(DeltaSchemaAdapterFactory {}))
.with_table_parquet_options(parquet_options);

// Add filter
if let Some(predicate) = logical_filter {
    if config.enable_parquet_pushdown {
        exec_plan_builder = exec_plan_builder.with_predicate(predicate);
    }
};

New code should use FileScanConfig to build the appropriate DataSourceExec:

let mut file_source = ParquetSource::new(parquet_options)
    .with_schema_adapter_factory(Arc::new(DeltaSchemaAdapterFactory {}));

// Add filter
if let Some(predicate) = logical_filter {
    if config.enable_parquet_pushdown {
        file_source = file_source.with_predicate(predicate);
    }
};

let file_scan_config = FileScanConfig::new(
    self.log_store.object_store_url(),
    file_schema,
    Arc::new(file_source),
)
.with_statistics(stats)
.with_projection(self.projection.cloned())
.with_limit(self.limit)
.with_table_partition_cols(table_partition_cols);

// Build the actual scan like this
parquet_scan: file_scan_config.build(),

`datafusion-cli` no longer automatically unescapes strings¶

datafusion-cli previously would incorrectly unescape string literals (see ticket for more details).

To escape ' in SQL literals, use '':

> select 'it''s escaped';
+----------------------+
| Utf8("it's escaped") |
+----------------------+
| it's escaped         |
+----------------------+
1 row(s) fetched.

To include special characters (such as newlines via \n) you can use an E literal string. For example

> select 'foo\nbar';
+------------------+
| Utf8("foo\nbar") |
+------------------+
| foo\nbar         |
+------------------+
1 row(s) fetched.
Elapsed 0.005 seconds.

Changes to array scalar function signatures¶

DataFusion 46 has changed the way scalar array function signatures are declared. Previously, functions needed to select from a list of predefined signatures within the ArrayFunctionSignature enum. Now the signatures can be defined via a Vec of psuedo-types, which each correspond to a single argument. Those psuedo-types are the variants of the ArrayFunctionArgument enum and are as follows:

Array: An argument of type List/LargeList/FixedSizeList. All Array arguments must be coercible to the same type.
Element: An argument that is coercible to the inner type of the Array arguments.
Index: An Int64 argument.

Each of the old variants can be converted to the new format as follows:

TypeSignature::ArraySignature(ArrayFunctionSignature::ArrayAndElement):

TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
    arguments: vec![ArrayFunctionArgument::Array, ArrayFunctionArgument::Element],
    array_coercion: Some(ListCoercion::FixedSizedListToList),
});

TypeSignature::ArraySignature(ArrayFunctionSignature::ElementAndArray):

TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
    arguments: vec![ArrayFunctionArgument::Element, ArrayFunctionArgument::Array],
    array_coercion: Some(ListCoercion::FixedSizedListToList),
});

TypeSignature::ArraySignature(ArrayFunctionSignature::ArrayAndIndex):

TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
    arguments: vec![ArrayFunctionArgument::Array, ArrayFunctionArgument::Index],
    array_coercion: None,
});

TypeSignature::ArraySignature(ArrayFunctionSignature::ArrayAndElementAndOptionalIndex):

TypeSignature::OneOf(vec![
    TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
        arguments: vec![ArrayFunctionArgument::Array, ArrayFunctionArgument::Element],
        array_coercion: None,
    }),
    TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
        arguments: vec![
            ArrayFunctionArgument::Array,
            ArrayFunctionArgument::Element,
            ArrayFunctionArgument::Index,
        ],
        array_coercion: None,
    }),
]);

TypeSignature::ArraySignature(ArrayFunctionSignature::Array):

TypeSignature::ArraySignature(ArrayFunctionSignature::Array {
    arguments: vec![ArrayFunctionArgument::Array],
    array_coercion: None,
});

Alternatively, you can switch to using one of the following functions which take care of constructing the TypeSignature for you:

Signature::array_and_element
Signature::array_and_element_and_optional_index
Signature::array_and_index
Signature::array

DataFusion Query Optimizer

Introduction

Upgrade Guides¶

DataFusion 49.0.0¶

datafusion.execution.collect_statistics now defaults to true¶

Metadata is now represented by FieldMetadata¶

New datafusion.execution.spill_compression configuration option¶

DataFusion 48.0.0¶

Expr::Literal has optional metadata¶

Expr::WindowFunction is now Boxed¶

The VARCHAR SQL type is now represented as Utf8View in Arrow¶

ListingOptions default for collect_stat changed from true to false¶

Processing FieldRef instead of DataType for user defined functions¶

Physical Expression return Field¶

FileFormat::supports_filters_pushdown replaced with FileSource::try_pushdown_filters¶

ParquetExec, AvroExec, CsvExec, JsonExec Removed¶

PartitionedFile added as an argument to the FileOpener trait¶

DataFusion 47.0.0¶

Upgrades to arrow-rs and arrow-parquet 55.0.0 and object_store 0.12.0¶

DisplayFormatType::TreeRender¶

Removed Deprecated APIs¶

FileScanConfig –> FileScanConfigBuilder¶

DataFusion 46.0.0¶

Use invoke_with_args instead of invoke() and invoke_batch()¶

ParquetExec, AvroExec, CsvExec, JsonExec deprecated¶

Cookbook: Changes to ParquetExecBuilder¶

Cookbook: Changes to ParquetExecBuilder¶

datafusion-cli no longer automatically unescapes strings¶

Changes to array scalar function signatures¶

DataFusion `49.0.0`¶

`datafusion.execution.collect_statistics` now defaults to `true`¶

Metadata is now represented by `FieldMetadata`¶

New `datafusion.execution.spill_compression` configuration option¶

DataFusion `48.0.0`¶

`Expr::Literal` has optional metadata¶

`Expr::WindowFunction` is now `Box`ed¶

The `VARCHAR` SQL type is now represented as `Utf8View` in Arrow¶

`ListingOptions` default for `collect_stat` changed from `true` to `false`¶

Processing `FieldRef` instead of `DataType` for user defined functions¶

Physical Expression return `Field`¶

`FileFormat::supports_filters_pushdown` replaced with `FileSource::try_pushdown_filters`¶

`ParquetExec`, `AvroExec`, `CsvExec`, `JsonExec` Removed¶

`PartitionedFile` added as an argument to the `FileOpener` trait¶

DataFusion `47.0.0`¶

Upgrades to `arrow-rs` and `arrow-parquet` 55.0.0 and `object_store` 0.12.0¶

`DisplayFormatType::TreeRender`¶

`FileScanConfig` –> `FileScanConfigBuilder`¶

DataFusion `46.0.0`¶

Use `invoke_with_args` instead of `invoke()` and `invoke_batch()`¶

`ParquetExec`, `AvroExec`, `CsvExec`, `JsonExec` deprecated¶

Cookbook: Changes to `ParquetExecBuilder`¶

Cookbook: Changes to `ParquetExecBuilder`¶

`datafusion-cli` no longer automatically unescapes strings¶