[AVRO-2488] GenericData.validate(...) has inconsistent behaviour with unknown datum

• Type: Bug
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Component/s: java
• Assignee: Unassigned
• Reporter: Ryan Skraba
• Created: 29/Jul/19 09:46
• Updated: 25/Jul/23 09:29

Description

I noticed this when working with AVRO-2642.  An unknown datum validates differently whether its in a UNION or not.

When using the schema from that JIRA with a GenericRecord:

{
  "type": "record",
  "name": "Lifetime",
  "fields": [
    {"name": "name", "type": "string"},
    {"name": "birth",
     "type": {"type": "string", "java-class": "java.util.Date"}},
    {"name": "death",
     "type": ["null", 
              {"type": "string", "java-class": "java.util.Date"}],
      "default": null
    }
  ]
}

The following code validates as false:

GenericRecord r = new GenericData.Record(theAboveSchema);
r.put("name", "William Shakespeare");
r.put("birth", new Date(-12801286800000L));
boolean isValid = GenericData.get().validate(r.getSchema(), r); // false

Fair enough – a specific record with that schema is valid, but *java-class* is a SpecificData concept that is unknown to GenericData.

The following code, however throws an exception:

GenericRecord r = new GenericData.Record(theAboveSchema);
r.put("name", "William Shakespeare");
r.put("birth", new Date(-12801286800000L));
r.put("death", new Date(-11161414800000L));
boolean isValid = GenericData.get().validate(r.getSchema(), r); 
// AvroRuntimeException: Unknown datum type java.util.Date

I would expect the two records to either be equally false (preferred) or to both throw an unknown datum type.

Issue Links

• is related to: AVRO-2462 Specific record hashCode() with custom stringable java-class throws exception