[AVRO-612] Schema.toString() strips out field docs

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.3.3
• Fix Version/s: 1.4.0
• Component/s: java
• Assignee: Doug Cutting
• Reporter: David Rosenstrauch
• Created: 13/Aug/10 00:01
• Updated: 13/Dec/11 19:06
• Resolved: 18/Aug/10 21:36

Description

Although avro can successfully parse schema text that contains a "doc" on a Schema.Field, when a Schema containing a field doc is serialized (via Schema.toString()) the doc does not get written.

The following JUnit test case demonstrates this problem:

TestAvroFieldDocSerialization.java

import junit.framework.TestCase;

import org.apache.avro.Schema;

public class TestAvroFieldDocSerialization extends TestCase {

	public void testAvroFieldDocSerialization() {
		String schemaStr =
			"{"+
			"	\"name\": \"Rec\","+
			"	\"type\": \"record\","+
			"	\"fields\" : ["+
			"		{\"name\": \"f\", \"type\": \"int\", \"doc\": \"test\"}"+
			"	]"+
			"}";
		Schema schema = Schema.parse(schemaStr);
		verifyFieldDoc(schema);

		schemaStr = schema.toString();
		schema = Schema.parse(schemaStr);
		verifyFieldDoc(schema);
	}

	private void verifyFieldDoc(Schema schema) {
		Schema.Field field = schema.getField("f");
		assertEquals("test", field.doc());
	}
}

Note that the first call to verifyFieldDoc() succeeds, while the second one fails. They should both succeed (in my opinion).