jdk11bug导致DefaultRheaIterator出现NullPointerException，是否考虑规避 #990
Closed
Description
Describe the bug
java.lang.NullPointerException
at com.alipay.sofa.jraft.util.BytesUtil$LexicographicByteArrayComparator.compare(BytesUtil.java:118)
at com.alipay.sofa.jraft.util.BytesUtil$LexicographicByteArrayComparator.compare(BytesUtil.java:112)
at com.alipay.sofa.jraft.util.BytesUtil.compare(BytesUtil.java:95)
at com.alipay.sofa.jraft.rhea.client.DefaultRheaIterator.hasNext(DefaultRheaIterator.java:61)
at com.betalpha.barbeyond.kv.RheaKVStoreBackend$1.hasNext(RheaKVStoreBackend.java:87)Steps to reproduce
这个bug准确来说应该是jdk的，项目中使用的是11.0.15，存在这个问题，升级到jdk17后不存在这个问题，中间哪个版本修复的未知，如找到，忘告知。具体的现象如果将iterator转化为stream时，即使DefaultRheaIterator.hasNext()返回false，还会多调用一次。
复现步骤:
Iterator<Object> iterator = new Iterator<>() {
@Override
public boolean hasNext() {
System.out.println("call hasNext");
return false;
}
@Override
public Object next() {
return new Object();
}
};
Stream<Object> stream1 = StreamSupport
.stream(Spliterators.spliteratorUnknownSize(iterator, Spliterator.ORDERED), false)
.map(obj -> {return obj;}); //这行必须
Stream<Object> stream2 = StreamSupport
.stream(Spliterators.spliteratorUnknownSize(stream1.iterator(), Spliterator.ORDERED), false);
System.out.println(stream2.collect(Collectors.toList()));结果是：
call hasNext
call hasNext
[]将上述代码的iterator替换为DefaultRheaIterator即可复现NPE。
看下是否有必要兼容下这种情况
Minimal yet complete reproducer code (or GitHub URL to code)
Environment
- SOFAJRaft version: jraft-rheakv-core-1.3.11.jar
- JVM version (e.g.
java -version): 11.0.15
Metadata
Assignees
Labels
No labels