[
https://issues.apache.org/jira/browse/SPARK-5456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14386628#comment-14386628
]
Karthik Gorthi commented on SPARK-5456:
---------------------------------------
One workaround we followed is to convert all Decimal Datatype columns to
integer (truncated).
But, this is not possible when the application needs to connect to third-party
databases were the datatype is obviously not under our control.
So, this is a serious bug, IMHO, which prohibits using Spark in cases, unless
there is a workaround?
> Decimal Type comparison issue
> -----------------------------
>
> Key: SPARK-5456
> URL: https://issues.apache.org/jira/browse/SPARK-5456
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 1.2.0, 1.3.0
> Reporter: Kuldeep
>
> Not quite able to figure this out but here is a junit test to reproduce this,
> in JavaAPISuite.java
> {code:title=DecimalBug.java}
> @Test
> public void decimalQueryTest() {
> List<Row> decimalTable = new ArrayList<Row>();
> decimalTable.add(RowFactory.create(new BigDecimal("1"), new
> BigDecimal("2")));
> decimalTable.add(RowFactory.create(new BigDecimal("3"), new
> BigDecimal("4")));
> JavaRDD<Row> rows = sc.parallelize(decimalTable);
> List<StructField> fields = new ArrayList<StructField>(7);
> fields.add(DataTypes.createStructField("a",
> DataTypes.createDecimalType(), true));
> fields.add(DataTypes.createStructField("b",
> DataTypes.createDecimalType(), true));
> sqlContext.applySchema(rows.rdd(),
> DataTypes.createStructType(fields)).registerTempTable("foo");
> Assert.assertEquals(sqlContext.sql("select * from foo where a >
> 0").collectAsList(), decimalTable);
> }
> {code}
> Fails with
> java.lang.ClassCastException: java.math.BigDecimal cannot be cast to
> org.apache.spark.sql.types.Decimal
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
