[FLINK-XXXXX][table] Refactor changelog inference program

[bvarghese1]

What is the purpose of the change

(For example: This pull request makes task deployment go through the blob server, rather than through RPC. That way we avoid re-transferring them on each deployment (during recovery).)

Brief change log

(for example:)

• The TaskInfo is stored in the blob store on job creation time as a persistent artifact
• Deployments RPC transmits only the blob storage reference
• TaskManagers retrieve the TaskInfo from the blob cache

Verifying this change

Please make sure both new and modified tests in this PR follow the conventions for tests defined in our code quality guide.

(Please pick either of the following options)

This change is a trivial rework / code cleanup without any test coverage.

(or)

This change is already covered by existing tests, such as (please describe tests).

(or)

This change added tests and can be verified as follows:

(example:)

• Added integration tests for end-to-end deployment with large payloads (100MB)
• Extended integration test for recovery after master (JobManager) failure
• Added test that validates that TaskInfo is transferred only once across recoveries
• Manually verified the change by running a 4 node cluster with 2 JobManagers and 4 TaskManagers, a stateful streaming program, and killing one JobManager and two TaskManagers during the execution, verifying that recovery happens correctly.

Does this pull request potentially affect one of the following parts:

• Dependencies (does it add or upgrade a dependency): (yes / no)
• The public API, i.e., is any changed class annotated with @Public(Evolving): (yes / no)
• The serializers: (yes / no / don't know)
• The runtime per-record code paths (performance sensitive): (yes / no / don't know)
• Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: (yes / no / don't know)
• The S3 file system connector: (yes / no / don't know)

Documentation

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

---

Was generative AI tooling used to co-author this PR?

• Yes (please specify the tool below)

[flinkbot]

CI report:

• 4aeb20c Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build

[gustavodemorais]

I agree the file's readability isn't great. We've now split the logic for update, modify, and delete across three files. When working in FlinkChangelogModeInferenceProgram I usually jump to one node and want all of its logic in one place, so I'd suggest other paths:

1. Convert to Java - definitely +1.
2. Move each node's logic into a registry (Map<Class, NodeHandler> or visitor-per-node). This keeps a node's logic together across traits and drops the instanceof ladder entirely, so "what does node X do" becomes a single lookup.
3. Give repeated branches a name - many nodes just forward or do full-delete-if-updates.
4. Name specific branches for readability, e.g. visitWithFallback(child, preferred, fallback) instead of Optional.or(() -> visit(child, fallback)).

Another option is splitting the instanceof ladder into one small private method per node (visitSink, visitJoin, visitCalc), so the top-level visit() becomes a short dispatch and each handler reads on its own. We already use this pattern in StreamNonDeterministicUpdatePlanVisitor. I lean toward the registry first, but glad to hear other opinions

[bvarghese1]

I agree the file's readability isn't great. We've now split the logic for update, modify, and delete across three files. When working in FlinkChangelogModeInferenceProgram I usually jump to one node and want all of its logic in one place, so I'd suggest other paths:

1. Convert to Java - definitely +1.
2. Move each node's logic into a registry (Map<Class, NodeHandler> or visitor-per-node). This keeps a node's logic together across traits and drops the instanceof ladder entirely, so "what does node X do" becomes a single lookup.
3. Give repeated branches a name - many nodes just forward or do full-delete-if-updates.
4. Name specific branches for readability, e.g. visitWithFallback(child, preferred, fallback) instead of Optional.or(() -> visit(child, fallback)).

Another option is splitting the instanceof ladder into one small private method per node (visitSink, visitJoin, visitCalc), so the top-level visit() becomes a short dispatch and each handler reads on its own. We already use this pattern in StreamNonDeterministicUpdatePlanVisitor. I lean toward the registry first, but glad to hear other opinions

Hi @gustavodemorais , yes that's definitely the plan :-) . But I want to do this refactoring in 2 phases. The first PR simply converts to Java with a few newly introduced classes like in this PR. The second PR will move each node into a registry (this is more involved and I did not want to complicate this initial PR).