Hi Shashank,
Scala case classes are treated as a special tuple type in Flink. If you
want to make a POJO out of it, just remove the "case" keyword and make
sure that the class is static (in the companion object).
I hope that helps.
Timo
Am 12/19/17 um 11:04 AM schrieb shashank agarwal:
HI,
I have upgraded flink from 1.3.2 to 1.4.0. I am using cassandra sink
in my scala application. Before sink, i was converting my scala
datastream to java stream and sinking in Cassandra. I have created
pojo class in scala liked that :
@SerialVersionUID(507L)
@Table(keyspace = "neofp", name = "order_detail")
case class OrderFinal(
@BeanProperty var order_name: String,
@BeanProperty var user: String )extends
Serializable
{
def this() {
this("NA", "NA",)
}
}
and this was working fine with sink. after upgrading to 1.4.0 it's
giving error "Query must not be null or empty."
After dig into the CassandraSink code, I have found it's treating it
as case class and running CassandraScalaProductSinkBuilder which
do sanityCheck of query existence.
So how I can create POJO class in scala so CassandraSink treats it
as CassandraPojoSinkBuilder?
For workaround now I have downgraded the only connector to 1.3.2
Thanks
Shashank
