You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Yang Jie (Jira)" <ji...@apache.org> on 2022/11/27 06:07:00 UTC
[jira] [Created] (SPARK-41276) Optimize constructor use of `StructType`
Yang Jie created SPARK-41276:
--------------------------------
Summary: Optimize constructor use of `StructType`
Key: SPARK-41276
URL: https://issues.apache.org/jira/browse/SPARK-41276
Project: Spark
Issue Type: Improvement
Components: MLlib, SQL
Affects Versions: 3.4.0
Reporter: Yang Jie
There are two main ways to construct `StructType`:
- Primary constructor
```scala
case class StructType(fields: Array[StructField])
```
- Use `Seq` as input constructor
```scala
def apply(fields: Seq[StructField]): StructType = StructType(fields.toArray)
```
These two construction methods are widely used in Spark, but the latter requires an additional collection conversion.
This pr changes the following 3 scenarios to use primary constructor to reduce one collection conversion:
1. For manually create `Seq` input scenes, change to use manually create `Array` input instead, for examaple:
https://github.com/apache/spark/blob/bcf03fe3f86a7230fd977c059b73a58554370d5d/mllib/src/main/scala/org/apache/spark/ml/image/ImageSchema.scala#L55-L63
2. For the scenario where 'toSeq' is added to create input for compatibility with Scala 2.13, directly call 'toArray' to instead, for example:
https://github.com/apache/spark/blob/bcf03fe3f86a7230fd977c059b73a58554370d5d/connector/avro/src/main/scala/org/apache/spark/sql/avro/SchemaConverters.scala#L108-L113
3. For scenes whose input is originally `Array`, remove the redundant `toSeq`, for example:
https://github.com/apache/spark/blob/bcf03fe3f86a7230fd977c059b73a58554370d5d/sql/catalyst/src/main/scala/org/apache/spark/sql/types/StructType.scala#L587-L592
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org