[Bug]: Certain nullable type resolved to not null coder after #32705

[Abacn]

What happened?

Caught by DataflowTemplates validation for Beam 2.61.0rc1: https://github.com/GoogleCloudPlatform/DataflowTemplates/pull/2014/checks?check_run_id=33057418339

Was able to bisect:

good: 20d0f6e

bad: c243491

To reproduce,

1. checkout Validate Beam 2.61.0rc1 GoogleCloudPlatform/DataflowTemplates#2014

2. on beam repo, after making any change, run

On Beam repo, after making any change (e.g. add log)

git cherry-pick c53a2017ba6e22816c8346fc8ce50142d30e3f81

./gradlew :sdks:java:core:publishToMavenLocal -Ppublishing

./gradlew :runners:direct-java:publishToMavenLocal -Ppublishing

3. On DataflowTemplate repo, run

mvn test -pl v2/dataplex -Dtest=DataplexBigQueryToGcsTest#testE2E_mainPathWithAllStepsEnabled

Issue Priority

Priority: 1 (data loss / total loss of function)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam YAML
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Infrastructure
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner

[Abacn]

A simple test: succeeded prior to the change, failing after the change:

package org.apache.beam.runners.direct;

import com.google.auto.value.AutoValue;
import java.io.Serializable;
import org.apache.beam.sdk.coders.DefaultCoder;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.SchemaCoder;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
import org.apache.beam.sdk.testing.TestPipeline;
import org.apache.beam.sdk.transforms.Create;
import org.apache.beam.sdk.transforms.MapElements;
import org.apache.beam.sdk.transforms.SerializableFunction;
import org.apache.beam.sdk.values.KV;
import org.apache.beam.sdk.values.TypeDescriptor;
import org.apache.beam.sdk.values.TypeDescriptors;
import org.checkerframework.checker.nullness.qual.Nullable;
import org.junit.Rule;
import org.junit.Test;
import org.junit.runner.RunWith;
import org.junit.runners.JUnit4;

@RunWith(JUnit4.class)
public class NullableEvaluationTest {
  @Rule
  public TestPipeline p = TestPipeline.create();

  @Test
  public void testConstruction() {
    // Either annotated with @Nullable or not does not matter
    p.apply(Create.of(KV.<String, @Nullable BigQueryTablePartition>of("123", null)))
        .apply("MapFileNames", MapElements.into(TypeDescriptors.kvs(
                TypeDescriptor.of(String.class),
                TypeDescriptor.of(BigQueryTablePartition.class)))
            .via(kv -> kv));
    p.run();
  }

  /** BigQuery table partition metadata. */
  @AutoValue
  @DefaultCoder(SchemaCoder.class)
  @DefaultSchema(AutoValueSchema.class)
  public abstract static class BigQueryTablePartition implements Serializable {

    public abstract String getPartitionName();

    public static Builder builder() {
      return new AutoValue_NullableEvaluationTest_BigQueryTablePartition.Builder();
    }

    /** Builder for {@link BigQueryTablePartition}. */
    @AutoValue.Builder
    public abstract static class Builder {
      public abstract Builder setPartitionName(String value);

      public abstract BigQueryTablePartition build();
    }
  }
}

[damccorm]

This has been fixed via #33133 - leaving open since we had to revert the PR from @reuvenlax and we probably want the right fix going forward here

[reuvenlax]

Can you explain what the bug was (other than some tests started failing)? It's not clear to me from this description.

[Abacn]

The bug was that it used to be able to encode elements of KV<KeyType, @Nullable ValueType> where ValueType uses a SchemaCoder, but now it cannot.