Github user anubhav100 commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/1990#discussion_r170196522
--- Diff:
integration/spark-common-test/src/test/scala/org/apache/carbondata/spark/testsuite/standardpartition/StandardPartitionTableQueryTestCase.scala
---
@@ -242,11 +242,13 @@ class StandardPartitionTableQueryTestCase extends
QueryTest with BeforeAndAfterA
}
-test("Creation of partition table should fail if the colname in table
schema and partition column is same even if both are case sensitive"){
- intercept[Exception]{
- sql("CREATE TABLE uniqdata_char2(name char,id int) partitioned by
(NAME char)stored by 'carbondata' ")
+ test("Creation of partition table should fail if the colname in table
schema and partition column is same even if both are case sensitive") {
+ val e = intercept[AnalysisException] {
+ sql("CREATE TABLE uniqdata_char2(name char,id int) partitioned by
(NAME char)stored by 'carbondata' ")
+ }
+ //TODO: error message is improper
+ assert(e.getMessage.contains("DataType char is not supported"))
--- End diff --
@xubo245 spark 2.1.0 supports creating table even if char data type is
given with out digits
scala> spark.sql("create table id(id char)");
18/02/23 14:12:23 WARN HiveMetaStore: Location:
file:/home/anubhav/Documents/phatak/spark-2.1/bin/spark-warehouse/id specified
for non-external table:id
res0: org.apache.spark.sql.DataFrame = []
but not in spark 2.2.1
cala> spark.sql("create table id(id char)");
18/02/23 14:22:05 WARN ObjectStore: Failed to get database global_temp,
returning NoSuchObjectException
org.apache.spark.sql.catalyst.parser.ParseException:
DataType char is not supported.(line 1, pos 19)
so to make sure that test case works fine for both spark version you can
change the create table ddl to
sql("CREATE TABLE uniqdata_char2(name char,id int) partitioned by (NAME
char(1))stored by 'carbondata' ")
the check for exception message
Operation not allowed: Partition columns should not be specified in the
schema: ["name"](line 1, pos 62)
otherwise it will be passed for spark 2.1 .0 but not for spark 2.2.1
---