Phase 24: Restore == comparison with HOODIE_METADATA_SCHEMA in HoodieMetadataPayload

[voonhous]

Task Description

Restore .equals comparison to use == to restore performance.

Before migration in phase 5:

• When schemas were read from log file headers, they went through AvroSchemaCache.intern(Schema)
• AvroSchemaCache is keyed by Schema (using Schema.equals()), so even if a new Schema object was parsed, intern() would return the same cached reference as HOODIE_METADATA_SCHEMA

After migration in phase 5:

• Schemas are parsed into HoodieSchema via HoodieSchema.parse()
• HoodieSchemaCache.intern(HoodieSchema) caches the HoodieSchema wrapper
• When .toAvroSchema() is called, it returns the underlying Avro Schema which was created by avroParser.parse()
• this is a NEW object, not the same reference as HOODIE_METADATA_SCHEMA

During this transitionary PRs, we have the following mismatch:

• AvroSchemaCache ensured Avro Schema reference reuse
• HoodieSchemaCache ensures HoodieSchema wrapper reuse, but the wrapped Avro Schema is still a different object

The .equals() is a temporary bridge during the migration.

Once HoodieRecordPayload.getInsertValue() accepts HoodieSchema instead of Schema, the == comparison can be restored.

Details of context can be found in this comment:
#14340 (comment)

What needs to be done:
Restore .equals comparison to use == to restore performance in HoodieMetadataPayload when checking against HOODIE_METADATA_SCHEMA after 1Avro.SchematoHoodieSchemamigration forHoodieRecordPayload` .

Why this task is needed:
Restore O(1) comparison

Task Type

Code improvement/refactoring

Related Issues

Parent feature issue: (if applicable )
Related issues:
NOTE: Use Relationships button to add parent/blocking issues after issue is created.