openinx commented on a change in pull request #1936:
URL: https://github.com/apache/iceberg/pull/1936#discussion_r561515156
##########
File path:
flink/src/test/java/org/apache/iceberg/flink/TestFlinkTableSource.java
##########
@@ -685,4 +782,60 @@ public void testSqlParseError() {
AssertHelpers.assertThrows("The NaN is not supported by flink now. ",
NumberFormatException.class, () -> sql(sqlParseErrorLTE));
}
+
+ /**
+ * The sql can be executed in both streaming and batch mode, in order to get
the parallelism, we convert the flink
+ * Table to flink DataStream, so we only use streaming mode here.
+ *
+ * @throws TableNotExistException table not exist exception
+ */
+ @Test
+ public void testInferedParallelism() throws TableNotExistException {
+ Assume.assumeTrue("The execute mode should be streaming mode",
isStreamingJob);
Review comment:
> Just as you commented that it is best not to use the internal code of
flink, I think we should try to use public APIs to get information.
The comment that saying we'd better not use flink's Internal API because
that would introduce extra upgrade complexity (new flink version may breaks
those internal API so we iceberg have to adjust the codes, finally maintaining
different versions of flink will bring us a lot of burden).
Writing iceberg unit tests based on our iceberg's non-public ( we usually
use package-access ) method is OK because there's no extra burden .
----------------------------------------------------------------
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]
