JingsongLi opened a new pull request, #9039:
URL: https://github.com/apache/paimon/pull/9039
## What changed
- Replace `VersionedObjectSerializer` with independent manifest serializer
implementations.
- Hard-code the permanent on-disk format identifiers (`2`, `2`, and `1`)
next to comments explaining that additive nullable fields must not change them.
- Preserve the historical leading `_VERSION` field through a
manifest-specific schema helper.
- Add coverage for the fixed identifiers and old-reader/new-writer
compatibility.
- Clarify that `ObjectSerializer` byte serialization requires the same row
schema for deserialization.
## Why
Manifest files support additive schema evolution because older readers
ignore unknown fields. Treating the leading value as a mutable serializer
version makes compatible field additions easy to turn into unnecessary format
incompatibilities.
Each persistent serializer now owns a permanent format identifier directly,
removing the version-bump abstraction and making the compatibility rule
explicit at the write site.
## Compatibility
The on-disk row layout and identifiers are unchanged:
- `ManifestEntry`: `2`
- `ManifestFileMeta`: `2`
- `IndexManifestEntry`: `1`
The unannotated internal `VersionedObjectSerializer` Java superclass is
removed intentionally.
## Validation
```text
mvn -pl paimon-core -am -DfailIfNoTests=false -DwildcardSuites=none \
-Dtest=ManifestEntrySerializerTest,ManifestFileMetaSerializerTest,IndexManifestEntrySerializerTest,ManifestListTest#testCanReadOldMetaPaimon10+testOldReaderCanReadNewMetaPaimon10,IndexManifestFileHandlerTest#testNewIndexManifestReadableWithLegacySchema
test
```
Result: 13 tests passed. Checkstyle, Spotless, Enforcer, and reactor
compilation also passed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
