Bohdan Kazydub created DRILL-6993:
-------------------------------------
Summary: VARBINARY length is ignored on cast
Key: DRILL-6993
URL: https://issues.apache.org/jira/browse/DRILL-6993
Project: Apache Drill
Issue Type: Bug
Reporter: Bohdan Kazydub
Assignee: Bohdan Kazydub
{{VARBINARY}} precision is not set when casting to {{VARBINARY}} with specified
length.
For example, test case
{code}
String query = "select cast(r_name as varbinary(31)) as vb from
cp.`tpch/region.parquet`;"
MaterializedField field = new ColumnBuilder("vb",
TypeProtos.MinorType.VARBINARY)
.setMode(TypeProtos.DataMode.OPTIONAL)
.setWidth(31)
.build();
BatchSchema expectedSchema = new SchemaBuilder()
.add(field)
.build();
// Validate schema
testBuilder()
.sqlQuery(q)
.schemaBaseLine(expectedSchema)
.go();
{code}
will fail with
{code}
java.lang.Exception: Schema path or type mismatch for column #0:
Expected schema path: vb
Actual schema path: vb
Expected type: MajorType[minor_type: VARBINARY mode: OPTIONAL precision: 31
scale: 0]
Actual type: MajorType[minor_type: VARBINARY mode: OPTIONAL]
{code}
while for other types, like {{VARCHAR}}, it seems to work.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
