rdblue commented on a change in pull request #1882:
URL: https://github.com/apache/iceberg/pull/1882#discussion_r537685778
##########
File path:
flink/src/test/java/org/apache/iceberg/flink/source/TestFlinkInputFormat.java
##########
@@ -100,6 +100,33 @@ public void testNestedProjection() throws Exception {
assertRows(result, expected);
}
+ @Test
+ public void testBasicProjection() throws IOException {
+ Schema writeSchema = new Schema(
+ Types.NestedField.required(0, "id", Types.LongType.get()),
+ Types.NestedField.optional(1, "data", Types.StringType.get()),
+ Types.NestedField.optional(2, "time", Types.TimestampType.withZone())
+ );
+
+ Table table = catalog.createTable(TableIdentifier.of("default", "t"),
writeSchema);
+
+ List<Record> writeRecords = RandomGenericData.generate(writeSchema, 2, 0L);
+ new GenericAppenderHelper(table, fileFormat,
TEMPORARY_FOLDER).appendToTable(writeRecords);
+
+ TableSchema projectedSchema = TableSchema.builder()
+ .field("id", DataTypes.BIGINT())
+ .field("data", DataTypes.STRING())
+ .build();
+ List<Row> result =
runFormat(FlinkSource.forRowData().tableLoader(loader()).project(projectedSchema).buildFormat());
Review comment:
I like the fix in this PR. If an expected type is not present, then it
is fine to return null and to skip adding it to the field map in the struct
method. I'd rather do that than return a different reader. At least that way,
we don't create extra readers that won't be used anyway.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
