NicoK commented on a change in pull request #6705: [FLINK-10356][network] add
sanity checks to SpillingAdaptiveSpanningRecordDeserializer
URL: https://github.com/apache/flink/pull/6705#discussion_r252362102
##########
File path:
flink-runtime/src/test/java/org/apache/flink/runtime/io/network/api/serialization/SpanningRecordSerializationTest.java
##########
@@ -105,11 +119,235 @@ public void testHandleMixedLargeRecords() throws
Exception {
testSerializationRoundTrip(originalRecords, segmentSize);
}
+ /**
+ * Non-spanning, deserialization reads one byte too many and succeeds -
failure report comes
+ * from an additional check in {@link
SpillingAdaptiveSpanningRecordDeserializer}.
+ */
+ @Test
+ public void testHandleDeserializingTooMuchNonSpanning1() throws
Exception {
+ testHandleWrongDeserialization(
+ DeserializingTooMuch.getValue(),
+ 32 * 1024);
+ }
+
+ /**
+ * Non-spanning, deserialization reads one byte too many and fails.
+ */
+ @Test
+ public void testHandleDeserializingTooMuchNonSpanning2() throws
Exception {
+ testHandleWrongDeserialization(
+ DeserializingTooMuch.getValue(),
+ (serializedLength) -> serializedLength,
+ isA(IndexOutOfBoundsException.class));
Review comment:
This is the current behaviour (non-spanning accesses arrays and spanning
provides an input view that resembles file/stream accesses, each with the
respective exceptions in case of failures)...
Do you think we should should be more generic and tolerate any `IOException`
here? But then the cause may not be right...
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
