[
https://issues.apache.org/jira/browse/FLINK-3750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15241113#comment-15241113
]
ASF GitHub Bot commented on FLINK-3750:
---------------------------------------
Github user fpompermaier commented on a diff in the pull request:
https://github.com/apache/flink/pull/1885#discussion_r59712911
--- Diff:
flink-batch-connectors/flink-jdbc/src/main/java/org/apache/flink/api/java/io/jdbc/JDBCInputFormat.java
---
@@ -241,10 +264,31 @@ private void addValue(OUT reuse) throws IOException,
SQLException {
reuse.setField(resultSet.getDouble(pos + 1), pos);
break;
case java.sql.Types.DECIMAL:
-
reuse.setField(resultSet.getBigDecimal(pos + 1).doubleValue(), pos);
- break;
+ //TODO manage null fields
+ try {
+ if (o == null){
+
reuse.setField("", pos);
+ } else{
+
reuse.setField(resultSet.getBigDecimal(pos + 1).toPlainString(), pos);
--- End diff --
Because Flink cannot handle null values (unless you're using strings).
Indeed we need to decide how to overcome such limitation (for JDBC almost all
tables have null values that is bad to convert to 0.0 as in the Double case...).
> Make JDBCInputFormat a parallel source
> --------------------------------------
>
> Key: FLINK-3750
> URL: https://issues.apache.org/jira/browse/FLINK-3750
> Project: Flink
> Issue Type: Improvement
> Components: Batch
> Affects Versions: 1.0.1
> Reporter: Flavio Pompermaier
> Assignee: Flavio Pompermaier
> Priority: Minor
> Labels: connector, jdbc
>
> At the moment the batch JDBC InputFormat does not support parallelism
> (NonParallelInput). I'd like to remove such limitation
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
