diff --git a/common/src/main/scala/org/apache/comet/CometConf.scala b/common/src/main/scala/org/apache/comet/CometConf.scala index 49eb55479b..2439f3b58a 100644 --- a/common/src/main/scala/org/apache/comet/CometConf.scala +++ b/common/src/main/scala/org/apache/comet/CometConf.scala @@ -125,16 +125,14 @@ object CometConf extends ShimCometConf { val SCAN_AUTO = "auto" val COMET_NATIVE_SCAN_IMPL: ConfigEntry[String] = conf("spark.comet.scan.impl") - .category(CATEGORY_SCAN) + .category(CATEGORY_PARQUET) .doc( - "The implementation of Comet Native Scan to use. Available modes are " + + "The implementation of Comet's Parquet scan to use. Available scans are " + s"`$SCAN_NATIVE_DATAFUSION`, and `$SCAN_NATIVE_ICEBERG_COMPAT`. " + - s"`$SCAN_NATIVE_DATAFUSION` is a fully native implementation of scan based on " + - "DataFusion. " + - s"`$SCAN_NATIVE_ICEBERG_COMPAT` is the recommended native implementation that " + - "exposes apis to read parquet columns natively and supports complex types. " + - s"`$SCAN_AUTO` (default) chooses the best scan.") - .internal() + s"`$SCAN_NATIVE_DATAFUSION` is a fully native implementation, and " + + s"`$SCAN_NATIVE_ICEBERG_COMPAT` is a hybrid implementation that supports some " + + "additional features, such as row indexes and field ids. " + + s"`$SCAN_AUTO` (default) chooses the best available scan based on the scan schema.") .stringConf .transform(_.toLowerCase(Locale.ROOT)) .checkValues(Set(SCAN_NATIVE_DATAFUSION, SCAN_NATIVE_ICEBERG_COMPAT, SCAN_AUTO)) diff --git a/docs/source/contributor-guide/ffi.md b/docs/source/contributor-guide/ffi.md index b1a51ecb2a..c40c189e99 100644 --- a/docs/source/contributor-guide/ffi.md +++ b/docs/source/contributor-guide/ffi.md @@ -177,9 +177,10 @@ message Scan { #### When ownership is NOT transferred to native: -If the data originates from `native_comet` scan (deprecated, will be removed in a future release) or from -`native_iceberg_compat` in some cases, then ownership is not transferred to native and the JVM may re-use the -underlying buffers in the future. +If the data originates from a scan that uses mutable buffers (such as Iceberg scans using the [hybrid Iceberg reader]), +then ownership is not transferred to native and the JVM may re-use the underlying buffers in the future. + +[hybrid Iceberg reader]: https://datafusion.apache.org/comet/user-guide/latest/iceberg.html#hybrid-reader It is critical that the native code performs a deep copy of the arrays if the arrays are to be buffered by operators such as `SortExec` or `ShuffleWriterExec`, otherwise data corruption is likely to occur. diff --git a/docs/source/contributor-guide/parquet_scans.md b/docs/source/contributor-guide/parquet_scans.md index bbacff4d93..7df9394882 100644 --- a/docs/source/contributor-guide/parquet_scans.md +++ b/docs/source/contributor-guide/parquet_scans.md @@ -19,71 +19,60 @@ under the License. # Comet Parquet Scan Implementations -Comet currently has three distinct implementations of the Parquet scan operator. The configuration property -`spark.comet.scan.impl` is used to select an implementation. The default setting is `spark.comet.scan.impl=auto`, and -Comet will choose the most appropriate implementation based on the Parquet schema and other Comet configuration -settings. Most users should not need to change this setting. However, it is possible to force Comet to try and use -a particular implementation for all scan operations by setting this configuration property to one of the following -implementations. - -| Implementation | Description | -| ----------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `native_comet` | **Deprecated.** This implementation provides strong compatibility with Spark but does not support complex types. This is the original scan implementation in Comet and will be removed in a future release. | -| `native_iceberg_compat` | This implementation delegates to DataFusion's `DataSourceExec` but uses a hybrid approach of JVM and native code. This scan is designed to be integrated with Iceberg in the future. | -| `native_datafusion` | This experimental implementation delegates to DataFusion's `DataSourceExec` for full native execution. There are known compatibility issues when using this scan. | - -The `native_datafusion` and `native_iceberg_compat` scans provide the following benefits over the `native_comet` -implementation: - -- Leverages the DataFusion community's ongoing improvements to `DataSourceExec` -- Provides support for reading complex types (structs, arrays, and maps) -- Delegates Parquet decoding to native Rust code rather than JVM-side decoding -- Improves performance - -> **Note on mutable buffers:** Both `native_comet` and `native_iceberg_compat` use reusable mutable buffers -> when transferring data from JVM to native code via Arrow FFI. The `native_iceberg_compat` implementation uses DataFusion's native Parquet reader for data columns, bypassing Comet's mutable buffer infrastructure entirely. However, partition columns still use `ConstantColumnReader`, which relies on Comet's mutable buffers that are reused across batches. This means native operators that buffer data (such as `SortExec` or `ShuffleWriterExec`) must perform deep copies to avoid data corruption. -> See the [FFI documentation](ffi.md) for details on the `arrow_ffi_safe` flag and ownership semantics. - -The `native_datafusion` and `native_iceberg_compat` scans share the following limitations: - -- When reading Parquet files written by systems other than Spark that contain columns with the logical type `UINT_8` - (unsigned 8-bit integers), Comet may produce different results than Spark. Spark maps `UINT_8` to `ShortType`, but - Comet's Arrow-based readers respect the unsigned type and read the data as unsigned rather than signed. Since Comet - cannot distinguish `ShortType` columns that came from `UINT_8` versus signed `INT16`, by default Comet falls back to - Spark when scanning Parquet files containing `ShortType` columns. This behavior can be disabled by setting - `spark.comet.scan.unsignedSmallIntSafetyCheck=false`. Note that `ByteType` columns are always safe because they can - only come from signed `INT8`, where truncation preserves the signed value. -- No support for default values that are nested types (e.g., maps, arrays, structs). Literal default values are supported. -- No support for datetime rebasing detection or the `spark.comet.exceptionOnDatetimeRebase` configuration. When reading - Parquet files containing dates or timestamps written before Spark 3.0 (which used a hybrid Julian/Gregorian calendar), - the `native_comet` implementation can detect these legacy values and either throw an exception or read them without - rebasing. The DataFusion-based implementations do not have this detection capability and will read all dates/timestamps - as if they were written using the Proleptic Gregorian calendar. This may produce incorrect results for dates before - October 15, 1582. -- No support for Spark's Datasource V2 API. When `spark.sql.sources.useV1SourceList` does not include `parquet`, - Spark uses the V2 API for Parquet scans. The DataFusion-based implementations only support the V1 API, so Comet - will fall back to `native_comet` when V2 is enabled. - -The `native_datafusion` scan has some additional limitations: +Comet currently has two distinct implementations of the Parquet scan operator. + +| Scan Implementation | Notes | +| ----------------------- | ---------------------- | +| `native_datafusion` | Fully native scan | +| `native_iceberg_compat` | Hybrid JVM/native scan | + +The configuration property +`spark.comet.scan.impl` is used to select an implementation. The default setting is `spark.comet.scan.impl=auto`, which +currently always uses the `native_iceberg_compat` implementation. Most users should not need to change this setting. +However, it is possible to force Comet to use a particular implementation for all scan operations by setting +this configuration property to one of the following implementations. For example: `--conf spark.comet.scan.impl=native_datafusion`. + +The following features are not supported by either scan implementation, and Comet will fall back to Spark in these scenarios: + +- `ShortType` columns, by default. When reading Parquet files written by systems other than Spark that contain + columns with the logical type `UINT_8` (unsigned 8-bit integers), Comet may produce different results than Spark. + Spark maps `UINT_8` to `ShortType`, but Comet's Arrow-based readers respect the unsigned type and read the data as + unsigned rather than signed. Since Comet cannot distinguish `ShortType` columns that came from `UINT_8` versus + signed `INT16`, by default Comet falls back to Spark when scanning Parquet files containing `ShortType` columns. + This behavior can be disabled by setting `spark.comet.scan.unsignedSmallIntSafetyCheck=false`. Note that `ByteType` + columns are always safe because they can only come from signed `INT8`, where truncation preserves the signed value. +- Default values that are nested types (e.g., maps, arrays, structs). Literal default values are supported. +- Spark's Datasource V2 API. When `spark.sql.sources.useV1SourceList` does not include `parquet`, Spark uses the + V2 API for Parquet scans. The DataFusion-based implementations only support the V1 API. +- Spark metadata columns (e.g., `_metadata.file_path`) +- No support for Dynamic Partition Pruning (DPP) + +The following shared limitation may produce incorrect results without falling back to Spark: + +- No support for datetime rebasing detection or the `spark.comet.exceptionOnDatetimeRebase` configuration. When + reading Parquet files containing dates or timestamps written before Spark 3.0 (which used a hybrid + Julian/Gregorian calendar), dates/timestamps will be read as if they were written using the Proleptic Gregorian + calendar. This may produce incorrect results for dates before October 15, 1582. + +The `native_datafusion` scan has some additional limitations, mostly related to Parquet metadata. All of these +cause Comet to fall back to Spark. - No support for row indexes -- `PARQUET_FIELD_ID_READ_ENABLED` is not respected [#1758] -- There are failures in the Spark SQL test suite [#1545] -- Setting Spark configs `ignoreMissingFiles` or `ignoreCorruptFiles` to `true` is not compatible with Spark +- No support for reading Parquet field IDs +- No support for `input_file_name()`, `input_file_block_start()`, or `input_file_block_length()` SQL functions. + The `native_datafusion` scan does not use Spark's `FileScanRDD`, so these functions cannot populate their values. +- No support for `ignoreMissingFiles` or `ignoreCorruptFiles` being set to `true` -## S3 Support - -There are some differences in S3 support between the scan implementations. - -### `native_comet` (Deprecated) +The `native_iceberg_compat` scan has the following additional limitation that may produce incorrect results +without falling back to Spark: -> **Note:** The `native_comet` scan implementation is deprecated and will be removed in a future release. +- Some Spark configuration values are hard-coded to their defaults rather than respecting user-specified values. + This may produce incorrect results when non-default values are set. The affected configurations are + `spark.sql.parquet.binaryAsString`, `spark.sql.parquet.int96AsTimestamp`, `spark.sql.caseSensitive`, + `spark.sql.parquet.inferTimestampNTZ.enabled`, and `spark.sql.legacy.parquet.nanosAsLong`. See + [issue #1816](https://github.com/apache/datafusion-comet/issues/1816) for more details. -The `native_comet` Parquet scan implementation reads data from S3 using the [Hadoop-AWS module](https://hadoop.apache.org/docs/stable/hadoop-aws/tools/hadoop-aws/index.html), which -is identical to the approach commonly used with vanilla Spark. AWS credential configuration and other Hadoop S3A -configurations works the same way as in vanilla Spark. - -### `native_datafusion` and `native_iceberg_compat` +## S3 Support The `native_datafusion` and `native_iceberg_compat` Parquet scan implementations completely offload data loading to native code. They use the [`object_store` crate](https://crates.io/crates/object_store) to read data from S3 and @@ -95,7 +84,8 @@ continue to work as long as the configurations are supported and can be translat #### Additional S3 Configuration Options -Beyond credential providers, the `native_datafusion` implementation supports additional S3 configuration options: +Beyond credential providers, the `native_datafusion` and `native_iceberg_compat` implementations support additional +S3 configuration options: | Option | Description | | ------------------------------- | -------------------------------------------------------------------------------------------------- | @@ -108,7 +98,8 @@ All configuration options support bucket-specific overrides using the pattern `f #### Examples -The following examples demonstrate how to configure S3 access with the `native_datafusion` Parquet scan implementation using different authentication methods. +The following examples demonstrate how to configure S3 access with the `native_datafusion` and `native_iceberg_compat` +Parquet scan implementations using different authentication methods. **Example 1: Simple Credentials** @@ -140,11 +131,8 @@ $SPARK_HOME/bin/spark-shell \ #### Limitations -The S3 support of `native_datafusion` has the following limitations: +The S3 support of `native_datafusion` and `native_iceberg_compat` has the following limitations: 1. **Partial Hadoop S3A configuration support**: Not all Hadoop S3A configurations are currently supported. Only the configurations listed in the tables above are translated and applied to the underlying `object_store` crate. 2. **Custom credential providers**: Custom implementations of AWS credential providers are not supported. The implementation only supports the standard credential providers listed in the table above. We are planning to add support for custom credential providers through a JNI-based adapter that will allow calling Java credential providers from native code. See [issue #1829](https://github.com/apache/datafusion-comet/issues/1829) for more details. - -[#1545]: https://github.com/apache/datafusion-comet/issues/1545 -[#1758]: https://github.com/apache/datafusion-comet/issues/1758 diff --git a/docs/source/contributor-guide/roadmap.md b/docs/source/contributor-guide/roadmap.md index ce9c414163..6d99ee5456 100644 --- a/docs/source/contributor-guide/roadmap.md +++ b/docs/source/contributor-guide/roadmap.md @@ -51,20 +51,6 @@ with benchmarks that benefit from this feature like TPC-DS. This effort can be t [#3349]: https://github.com/apache/datafusion-comet/pull/3349 [#3510]: https://github.com/apache/datafusion-comet/issues/3510 -### Removing the native_comet scan implementation - -The `native_comet` scan implementation is now deprecated and will be removed in a future release ([#2186], [#2177]). -This is the original scan implementation that uses mutable buffers (which is incompatible with best practices around -Arrow FFI) and does not support complex types. - -Now that the default `auto` scan mode uses `native_iceberg_compat` (which is based on DataFusion's `DataSourceExec`), -we can proceed with removing the `native_comet` scan implementation, and then improve the efficiency of our use of -Arrow FFI ([#2171]). - -[#2186]: https://github.com/apache/datafusion-comet/issues/2186 -[#2171]: https://github.com/apache/datafusion-comet/issues/2171 -[#2177]: https://github.com/apache/datafusion-comet/issues/2177 - ## Ongoing Improvements In addition to the major initiatives above, we have the following ongoing areas of work: