[test](regression) Complete Paimon read/write P0 coverage

[Gabriel39]

What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary:

The Paimon P0 regression suite did not distinguish the four primary-key merge engines and did not verify that every unsupported Doris data-write shape leaves both data and snapshots unchanged. This PR:

• adds Parquet and ORC coverage for deduplicate, partial-update, aggregation, and first-row tables under automatic and forced-JNI reader routing;
• verifies dynamic-bucket cross-partition deduplication;
• covers rejected INSERT, INSERT SELECT, INSERT OVERWRITE, UPDATE, DELETE, and MERGE statements with pre/post row and snapshot invariants;
• records the Paimon CTAS metadata-atomicity defect as an isolated opt-in negative regression; and
• documents the Paimon/Iceberg read/write P0 coverage matrix and why no additional Iceberg P0 case is needed for the currently supported surface.

Release note

None

Check List (For Author)

• Test
  • Regression test
  • Unit Test
  • Manual test (add detailed scripts or steps below)
  • No need to test or manual test. Explain why:
    • This is a refactor/code format and no logic has been changed.
    • Previous test can cover this change.
    • No code files have been changed.
    • Other reason

Validation:

• Paimon merge-engine matrix and write-boundary suites: passed, 2/2.

• Existing Iceberg MERGE semantics suite: passed.

• Isolated Paimon CTAS negative suite: reproduced the expected metadata-atomicity failure and cleaned up the target table.

• FE and BE build: passed.

• Behavior changed:

  • No.
  • Yes.

• Does this need documentation?

  • No.
  • Yes.

Check List (For Reviewer who merge this PR)

• Confirm the release note
• Confirm test cases
• Confirm document
• Add branch pick label

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[Gabriel39]

run buildall

[Gabriel39]

/review

[github-actions]

Automated review conclusion: REQUEST_CHANGES — one P2 coverage-evidence defect; no P0/P1 blocker found.

The new merge-engine and write-boundary tests are otherwise internally consistent, but the coverage document points reader-parity maintainers to a suite whose advertised toggle no longer changes current-FE routing. The inline comment identifies the deterministic suite that should be cited instead.

Critical checkpoint conclusions:

• Goal and proof: the added suites exercise the intended Paimon merge semantics and unsupported-write boundary; the one remaining defect is the inaccurate JNI/native evidence mapping.
• Scope and clarity: the patch is focused on regression coverage and its coverage inventory.
• Concurrency/thread safety: no production concurrency is changed; the reviewed test concurrency and session isolation introduce no new shared-state defect.
• Error handling: the negative DML cases match concrete analyzer/command rejection paths before writer construction; the CTAS metadata leak is explicitly isolated as a gated known-bug reproducer.
• Memory safety and BE nullable handling: not applicable to these test/document-only changes.
• Lifecycle: catalogs and session variables are restored in finally blocks; no new production lifecycle is introduced.
• Configuration: only existing regression gates/session variables are used and restored; no runtime configuration is added.
• Compatibility and parallel paths: no API, storage-format, or rolling-upgrade contract changes; Parquet/ORC, automatic/forced-JNI, and the relevant write variants were traced.
• Conditional checks and comments: the opt-in CTAS gate and merge-engine comments match the current behavior.
• Test coverage: positive merge-engine and negative write-boundary cases are broad; reader parity already has an explicit route-asserting suite, but line 72 cites the stale non-proof.
• Expected outputs/data correctness: all 105 merge-matrix golden rows and 15 boundary-output rows were statically checked against their queries and expected semantics; ordering/scalar determinism is present.
• Observability: no production path is changed, so no new logs or metrics are required.
• Transaction/persistence and writes: the six unsupported writes reject before transaction/writer creation; no EditLog or persistent product state is changed by this patch.
• FE/BE variables: none are added; existing reader-type propagation was traced across FE and both BE scanner generations.
• Performance: fixtures are bounded and no production hot path changes.
• Other issues: suite counts and referenced paths were verified; no additional substantiated issue remained after three review rounds.

Validation was static only. Per the review-runner instructions, no build or test command was run. No additional user focus was provided.

[github-actions]

[P2] Cite the suite that actually verifies JNI/native routing

test_paimon_cpp_reader does not establish this row's parity on the current FE. Logical splits are explicitly tagged PAIMON_JNI, so BE ignores enable_paimon_cpp_reader; raw splits auto-route native in both halves. The false/true toggle therefore compares the same per-split routing and cannot prove a second reader. test_paimon_partition_pk_delete_refs already asserts JNI-only versus positive-native plans (including SplitStat types) for both Parquet and ORC. Please cite that explicit proof here, or add equivalent route assertions to the listed matrix/mixed-format suites instead of relying on the stale CPP test.

[Gabriel39]

run buildall

1

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.