ableegoldman commented on a change in pull request #6592:
URL: https://github.com/apache/kafka/pull/6592#discussion_r621676170
##########
File path:
clients/src/main/java/org/apache/kafka/common/serialization/ListSerializer.java
##########
@@ -77,21 +87,39 @@ public void configure(Map<String, ?> configs, boolean
isKey) {
}
}
+ private void serializeNullIndexList(final DataOutputStream out,
List<Inner> data) throws IOException {
+ List<Integer> nullIndexList = IntStream.range(0, data.size())
+ .filter(i -> data.get(i) == null)
+ .boxed().collect(Collectors.toList());
+ out.writeInt(nullIndexList.size());
+ for (int i : nullIndexList) out.writeInt(i);
+ }
+
@Override
public byte[] serialize(String topic, List<Inner> data) {
if (data == null) {
return null;
}
- final int size = data.size();
try (final ByteArrayOutputStream baos = new ByteArrayOutputStream();
final DataOutputStream out = new DataOutputStream(baos)) {
+ out.writeByte(serStrategy.ordinal()); // write serialization
strategy flag
+ if (serStrategy == SerializationStrategy.NULL_INDEX_LIST) {
+ serializeNullIndexList(out, data);
+ }
+ final int size = data.size();
out.writeInt(size);
for (Inner entry : data) {
- final byte[] bytes = inner.serialize(topic, entry);
- if (!isFixedLength) {
- out.writeInt(bytes.length);
+ if (entry == null) {
+ if (serStrategy == SerializationStrategy.NEGATIVE_SIZE) {
+ out.writeInt(Serdes.ListSerde.NEGATIVE_SIZE_VALUE);
+ }
+ } else {
+ final byte[] bytes = inner.serialize(topic, entry);
+ if (!isFixedLength || serStrategy ==
SerializationStrategy.NEGATIVE_SIZE) {
+ out.writeInt(bytes.length);
Review comment:
Just to clarify you mean don't expose this to the user at all, right?
That sounds completely fine to me. If there are enough people trying to
serialize lists of custom classes with all constant data size who want this
optimization exposed for general use, then someone will request the feature and
we can go back and add it in. Then we can debate what the API should look like
at that time, and keep things simple for now. Personally I suspect the vast
majority of non-primitive data types are not going to be constant size anyways.
Given the above, I think whether to track the strategy as an actual
`SerializationStrategy` enum vs a boolean flag becomes a matter of code style
and personal preference, since it's no longer exposed to the user. So it's up
to you whether you find the enum or the flag to be more readable or clean
--
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:
us...@infra.apache.org
