apache / apache/beam

Bug in RuntimeValueProvider JSON serialization

Open
#19,507 0 comments 0 reactions 0 assignees View on GitHub
bug core java P3
Dominant language
Java
Stars
8.7k
Forks
4.7k
Avg merge
1d 20h
Merged PRs (30d)
196

Description

Classes affected:
org.apache.beam.sdk.options.ValueProvider.Serializer
org.apache.beam.sdk.options.ValueProvider.Deserializer

The problem is that according to the JsonDeserializer documentation, the deserialize method isn't executed for null nodes:
( [https://static.javadoc.io/com.fasterxml.jackson.core/jackson-databind/2.9.6/com/fasterxml/jackson/databind/JsonDeserializer.html#deserialize(com.fasterxml.jackson.core.JsonParser,%20com.fasterxml.jackson.databind.DeserializationContext)](https://static.javadoc.io/com.fasterxml.jackson.core/jackson-databind/2.9.6/com/fasterxml/jackson/databind/JsonDeserializer.html#deserialize(com.fasterxml.jackson.core.JsonParser,%20com.fasterxml.jackson.databind.DeserializationContext) )
)
"Note that this method is never called for JSON null literal, and thus deserializers need (and should) not check for it."

If we serialize a RuntimeValueProvider, the isAccessible() will return false, so we call a writeNull(). During deserialization this isn't handled properly as mentioned and our deserialization will return null.

The end result is that getters with ValueProvider return values will return "null". AFAIK ValueProvider getters should be never null.

My guess is that either we should completely omit serializing RuntimeValueProviders, or during deserialization the proper runtime value provider should be created again - which requires more than just a simple "null" being present in the json.

Imported from Jira [BEAM-6963](https://issues.apache.org/jira/browse/BEAM-6963). Original Jira may contain additional context.
Reported by: bnemeth.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.