AVRO-4332: [java] Add support for enum default values in IDL serialization

[prygunov]

What is the purpose of the change

This pull request fixes a bug in IdlUtils.writeSchema where enum schemas with a default value would silently lose that default when serialized to IDL format. As a result, round-tripping a schema through IDL would drop the enum default.

Verifying this change

Added enumDefaultIsWrittenToIdl test in IdlUtilsTest that:

• Verifies an enum schema with a default value serializes correctly (output contains } = ACTIVE;)
• Verifies an enum schema without a default value is unaffected (output contains no =)

Documentation

• Does this pull request introduce a new feature? no
• If yes, how is the feature documented? not applicable

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adds support for emitting enum default values when serializing schemas to Avro IDL, with accompanying tests to validate the output.

Changes:

• Update IDL serialization to append = <default>; after enum definitions when an enum default is present.
• Add a unit test to verify enum defaults are (and are not) written to IDL output.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
lang/java/idl/src/main/java/org/apache/avro/idl/IdlUtils.java	Writes enum defaults into the generated IDL output when present on the schema.
lang/java/idl/src/test/java/org/apache/avro/idl/IdlUtilsTest.java	Adds coverage to assert enum default serialization behavior.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[prygunov]

@RyanSkraba can you take this to the next RC?)

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

Suppressed comments (1)

lang/java/idl/src/test/java/org/apache/avro/idl/IdlUtilsTest.java:117

• The PR description says the test also verifies that an enum schema without a default value is unaffected (no "=") but the current test only covers the "with default" case. To avoid regressions (e.g., always emitting "= ..." or a stray "="), add a negative assertion (or a second test) for an enum created without a default, or update the PR description to match what is actually tested.

  @Test
  public void enumDefaultIsWrittenToIdl() throws IOException {
    Schema withDefault = Schema.createEnum("Status", null, "naming", asList("ACTIVE", "INACTIVE"), "ACTIVE");

    StringWriter withDefaultWriter = new StringWriter();
    IdlUtils.writeIdlProtocol(withDefaultWriter, withDefault);

    assertTrue(withDefaultWriter.toString().contains("} = ACTIVE;"),
        "Enum with default should serialize default value");
  }

[iemejia]

Hi @prygunov, sorry for taking so long to review this — thanks for the fix and for your patience.

The change is correct and a clear improvement. The writer was asymmetric with IdlReader (which already parses enum defaults via exitEnumDefault), so any enum with a default was silently dropped on round-trip. The fix matches both the IDL grammar (RBrace defaultSymbol=enumDefault? → } = SYMBOL;) and the existing fixtures (status_schema.avdl, simple.avdl). I applied it locally and confirmed a full write → parse round-trip now preserves getEnumDefault(), with no change for enums that have no default.

A couple of things before we can merge:

1. JIRA ticket + title. Per our contribution guidelines, PRs need an associated JIRA issue and the title should follow AVRO-XXXX: [java] .... Could you please create a ticket at https://issues.apache.org/jira/projects/AVRO/issues/ and update the PR title accordingly (e.g. AVRO-XXXX: [java] Add support for enum default values in IDL serialization)?

2. Strengthen the test. The current test only covers the positive case, though the PR description also mentions the "no default" case. Could you add:

   • a negative assertion that an enum without a default emits no =, and
   • a real round-trip assertion that proves the reported bug is fixed — write to IDL, parse it back, and check the default survives:

   IdlReader reader = new IdlReader();
   Schema roundTripped;
   try (InputStream in = new ByteArrayInputStream(withDefaultWriter.toString().getBytes(StandardCharsets.UTF_8))) {
     roundTripped = reader.parse(in).getNamedSchemas().get("naming.Status");
   }
   assertEquals("ACTIVE", roundTripped.getEnumDefault());

Thanks again!

[iemejia]

Thanks @prygunov — the enum-default fix is correct. I pushed some changes directly to finish it up (force-pushed, so reset your local copy):

1. Squashed your 3 commits into one under your authorship with a proper AVRO-4332 message (tree unchanged).

2. Added a maintainer commit fixing pre-existing breakage your PR surfaced. The IdlUtilsTest class never actually ran in CI — Surefire only includes **/Test** (names starting with Test), so it's been dormant since it was added. Your new tests wouldn't have run either (not your fault — you followed the file's convention). Renaming it to TestIdlUtils uncovered latent bugs I fixed:

• Real bug in IdlUtils: the byte[] serializer discarded its output instead of writing to the generator, so bytes serialized to "".
• The callToJson test helper had the same discard bug.
• The validateHappyFlow* fixtures were in the wrong package (unreachable → NPE); moved and regenerated them.
• Dropped obsolete union-unwrapping and fixed the map/collection separator expectations.

The idl module now runs 39 green tests (your 4 enum tests + 18 previously hidden).

[iemejia]

LGTM