You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@drill.apache.org by "Arina Ielchiieva (JIRA)" <ji...@apache.org> on 2018/07/19 10:48:00 UTC

[jira] [Updated] (DRILL-6620) Remove deprecated parquet methods usage from ParquetMetaStatCollector class

     [ https://issues.apache.org/jira/browse/DRILL-6620?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Arina Ielchiieva updated DRILL-6620:
------------------------------------
    Description: 
After upgrade to parquet lib 1.10 there are many parquet methods that are marked as deprecated. Need to update Drill code to use newer methods, especially in 
https://github.com/apache/drill/blob/c6549e58859397c88cb1de61b4f6eee52a07ed0c/exec/java-exec/src/main/java/org/apache/drill/exec/store/parquet/stat/ParquetMetaStatCollector.java

Making such changes would require to change the way how we create parquet statistics from Drill's {{ColumnMetadata}} class. Since currently it only stores {{PrimitiveType.PrimitiveTypeName}} info, we might need to change {{ColumnMetadata}}  to hold more information to move to new way of creating stats.
{code}
  /**
   * Returns the typed statistics object based on the passed type parameter
   * @param type PrimitiveTypeName type of the column
   * @return instance of a typed statistics class
   * @deprecated Use {@link #createStats(Type)} instead
   */
  @Deprecated
  public static Statistics getStatsBasedOnType(PrimitiveTypeName type) {
...
{code}

Also consider renaming {{ColumnMetadata.getNulls}} method to {{getNumNulls}} to be consistent with parquet stats class naming.

  was:
After upgrade to parquet lib 1.10 there are many parquet methods that are marked as deprecated. Need to update Drill code to use newer methods, especially in 
https://github.com/apache/drill/blob/c6549e58859397c88cb1de61b4f6eee52a07ed0c/exec/java-exec/src/main/java/org/apache/drill/exec/store/parquet/stat/ParquetMetaStatCollector.java

Making such changes would require to change the way how we create parquet statistics from Drill's {{ColumnMetadata}} class. Since currently it only stores {{PrimitiveType.PrimitiveTypeName}} info, we might need to change {{ColumnMetadata}}  to hold more information to move to new way of creating stats.
{code}
  /**
   * Returns the typed statistics object based on the passed type parameter
   * @param type PrimitiveTypeName type of the column
   * @return instance of a typed statistics class
   * @deprecated Use {@link #createStats(Type)} instead
   */
  @Deprecated
  public static Statistics getStatsBasedOnType(PrimitiveTypeName type) {
...
{code}


> Remove deprecated parquet methods usage from ParquetMetaStatCollector class
> ---------------------------------------------------------------------------
>
>                 Key: DRILL-6620
>                 URL: https://issues.apache.org/jira/browse/DRILL-6620
>             Project: Apache Drill
>          Issue Type: Task
>    Affects Versions: 1.14.0
>            Reporter: Arina Ielchiieva
>            Priority: Major
>
> After upgrade to parquet lib 1.10 there are many parquet methods that are marked as deprecated. Need to update Drill code to use newer methods, especially in 
> https://github.com/apache/drill/blob/c6549e58859397c88cb1de61b4f6eee52a07ed0c/exec/java-exec/src/main/java/org/apache/drill/exec/store/parquet/stat/ParquetMetaStatCollector.java
> Making such changes would require to change the way how we create parquet statistics from Drill's {{ColumnMetadata}} class. Since currently it only stores {{PrimitiveType.PrimitiveTypeName}} info, we might need to change {{ColumnMetadata}}  to hold more information to move to new way of creating stats.
> {code}
>   /**
>    * Returns the typed statistics object based on the passed type parameter
>    * @param type PrimitiveTypeName type of the column
>    * @return instance of a typed statistics class
>    * @deprecated Use {@link #createStats(Type)} instead
>    */
>   @Deprecated
>   public static Statistics getStatsBasedOnType(PrimitiveTypeName type) {
> ...
> {code}
> Also consider renaming {{ColumnMetadata.getNulls}} method to {{getNumNulls}} to be consistent with parquet stats class naming.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)