fix two flaky tests

[arianacai1997]

There are two flaky tests. They are because of the order of twin and samplePair. By running nondex, the order of one and two is nondeterministic. Therefore, I fixed them by adding another possibility

[yawkat]

Why do you think this test is flaky? The serialization order for Pair should be deterministic. If it's not, that seems like a bug.

[arianacai1997]

When I ran nondex, I got information like \ DeserializerTest.primitivePairs:659 expected:<{"[one":"f","two":"e]"}> but was:<{"[two":"e","one":"f]"}> \ DeserializerTest.twin:673 expected:<{"[one":"n","two":"h]"}> but was:<{"[two":"h","one":"n]"}>\ By looking at the debug file, it seems that mapper is an ObjectMapper with nondeterministic order.\ Here is a part of the traceback"":

java.lang.Class.getDeclaredFields(Class.java:1916)
com.fasterxml.jackson.databind.introspect.AnnotatedFieldCollector._findFields(AnnotatedFieldCollector.java:73)
com.fasterxml.jackson.databind.introspect.AnnotatedFieldCollector._findFields(AnnotatedFieldCollector.java:71)
com.fasterxml.jackson.databind.introspect.AnnotatedFieldCollector.collect(AnnotatedFieldCollector.java:48)
com.fasterxml.jackson.databind.introspect.AnnotatedFieldCollector.collectFields(AnnotatedFieldCollector.java:43)
com.fasterxml.jackson.databind.introspect.AnnotatedClass._fields(AnnotatedClass.java:371)
com.fasterxml.jackson.databind.introspect.AnnotatedClass.fields(AnnotatedClass.java:343)
com.fasterxml.jackson.databind.introspect.POJOPropertiesCollector._addFields(POJOPropertiesCollector.java:493)
com.fasterxml.jackson.databind.introspect.POJOPropertiesCollector.collectAll(POJOPropertiesCollector.java:421)
com.fasterxml.jackson.databind.introspect.POJOPropertiesCollector.getJsonValueAccessor(POJOPropertiesCollector.java:270)
com.fasterxml.jackson.databind.introspect.BasicBeanDescription.findJsonValueAccessor(BasicBeanDescription.java:258)
com.fasterxml.jackson.databind.ser.BasicSerializerFactory.findSerializerByAnnotations(BasicSerializerFactory.java:391)
com.fasterxml.jackson.databind.ser.BeanSerializerFactory._createSerializer2(BeanSerializerFactory.java:224)
com.fasterxml.jackson.databind.ser.BeanSerializerFactory.createSerializer(BeanSerializerFactory.java:173)
com.fasterxml.jackson.databind.SerializerProvider._createUntypedSerializer(SerializerProvider.java:1495)
com.fasterxml.jackson.databind.SerializerProvider._createAndCacheUntypedSerializer(SerializerProvider.java:1443)
com.fasterxml.jackson.databind.SerializerProvider.findValueSerializer(SerializerProvider.java:544)
com.fasterxml.jackson.databind.SerializerProvider.findTypedValueSerializer(SerializerProvider.java:822)
com.fasterxml.jackson.databind.ser.DefaultSerializerProvider.serializeValue(DefaultSerializerProvider.java:308)
com.fasterxml.jackson.databind.ObjectMapper._writeValueAndClose(ObjectMapper.java:4568)
com.fasterxml.jackson.databind.ObjectMapper.writeValueAsString(ObjectMapper.java:3821)
com.fasterxml.jackson.datatype.eclipsecollections.DeserializerTest.twin(DeserializerTest.java:673)

[yawkat]

Do you have a setup where this reproduces without nondex, whatever that is? Because the test cases aren't flaky for me locally.

[arianacai1997]

Do you have a setup where this reproduces without nondex, whatever that is? Because the test cases aren't flaky for me locally. I used these commands. But you might need to adapt the server to your VM.

mvn install -pl eclipse-collections -am -DskipTests | tee mvn-install.log
mvn -pl eclipse-collections test -Dtest=com.fasterxml.jackson.datatype.eclipsecollections.DeserializerTest#primitivePairs
mvn -pl eclipse-collections edu.illinois:nondex-maven-plugin:1.1.2:nondex -DnondexRuns=10 -Dtest=com.fasterxml.jackson.datatype.eclipsecollections.DeserializerTest#primitivePairs

[yawkat]

Look I get that it's erroring with nondex, but is it actually breaking in the real world? Outside of nondex?

[arianacai1997]

Outside

Sorry I didn't get what you mean. Nondex is a tool for order-related flaky test detection. Maybe you can refer to the GitHub for installation. https://github.com/TestingResearchIllinois/NonDex.

[yawkat]

Look. I understand that nondex reports this as an issue. However I will not consider this a bug unless you can show a case where the test fails without nondex being involved.

If I understand the issue correctly, then I wouldn't rule it out that this can happen in the real world. But I will not accept a bug that apparently exclusively appears in a runtime that was specifically designed to trigger that bug, and will never happen in the real world.

[arianacai1997]

Thanks for the explanation. I'll look at the issue and update it accordingly.

[cowtowncoder]

There seems to be a class of reported potential (but not necessarily actual, although could be) issues. I'll close this until such time as there's something to actually reproduce an issue, or at least verify fix we might try.