Aman Sinha created DRILL-2222:
---------------------------------

             Summary: For Parquet data check for schema change when we read 
footers the first time
                 Key: DRILL-2222
                 URL: https://issues.apache.org/jira/browse/DRILL-2222
             Project: Apache Drill
          Issue Type: Improvement
          Components: Query Planning & Optimization
    Affects Versions: 0.7.0
            Reporter: Aman Sinha
            Assignee: Jinfeng Ni


For parquet data, we are already reading the footers for row count information 
during the planning phase.  It would make sense to also check for schema change 
as we read each file's footer and throw an error early, otherwise we could fail 
in unexpected ways later during execution.  For example, here's a run-time 
error:  
{code}
 "java.lang.IllegalStateException: Failure while reading vector.  Expected 
vector class of org.apache.drill.exec.vector.NullableDecimal18Vector but was 
holding vector class 
org.apache.drill.exec.vector.NullableDecimal28SparseVector."
{code}

The reason for this error is that some data files have the column defined as 
decimal(15, 6) while others have decimal(19, 6).  



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to