[SkipRecovery] Re-enable Parquet V2 encoding fixtures [fast-ut] [databricks]

[wjxiz1992]

JaCoCo production line coverage: +0 lines (measured: delta-lake, shuffle-plugin, sql-plugin, udf-compiler; shim 350, vs nightly b19; delta-lake report group is stub-only; Iceberg implementation not loaded/applicable)

Contributes to NVIDIA/cudf#13501.

Description

Five Parquet V2 encoding fixtures remained xfailed after cuDF added the corresponding reader support in NVIDIA/cudf#13637, NVIDIA/cudf#14101, and NVIDIA/cudf#14590. The resolved cudf-spark JNI artifact embeds cuDF revision c2c4e85881a52e93e042a59cf88b7633147760c2, which contains all three fixes.

This change removes only those five fixture entries from _xfail_files. The existing test_parquet_testing_valid_files test continues to compare CPU and GPU results under both native and Java reader configurations, recovering 10 test nodes. Other issue-linked xfails in the shared mapping are unchanged.

Validation:

• Spark 3.5 distribution and integration-test build: all 16 reactor modules succeeded; BUILD SUCCESS.
• Forced pre-edit unmasked run of the five fixtures: 10 passed, 86 deselected, 12 warnings in 7.90s.
• Focused post-edit run without --runxfail: 10 passed, 86 deselected, 12 warnings in 8.91s.
• Complete parquet_testing_test.py: 84 passed, 4 xfailed, 8 xpassed, 90 warnings in 36.49s.
• All 10 recovered JUnit nodes passed; the plans contained 30 GpuFileGpuScan occurrences and no CPU-fallback markers.
• Post-edit exemption scan: zero [FEA] Support V2 encodings in Parquet reader and writer cudf#13501 matches; the five parameter cases are recovered while unrelated shared-map trackers remain.

AI assistance: The change was prepared with Codex assistance and reviewed by the author at commit ab12239d371d07abdd300422230a81b3eebec0ca before submission. The PR description was prepared with Codex assistance.

Checklists

Documentation

• Updated for new or modified user-facing features or behaviors
• No user-facing change

Testing

• Added or modified tests to cover new code paths
• Covered by existing tests
  (Please provide the names of the existing tests in the PR description.)
• Not required

Performance

• Tests ran and results are added in the PR description
• Issue filed with a link in the PR description
• Not required

[Copilot]

Pull request overview

This pull request re-enables five Parquet V2 “delta encoding” fixtures in the parquet-testing integration suite by removing their entries from the shared _xfail_files mapping now that upstream cuDF support is available. This restores GPU-vs-CPU parity coverage for those datasets under both native and Java parquet reader configurations.

Changes:

• Removed five delta_* parquet-testing fixtures from _xfail_files so they are no longer marked xfail.
• Restored execution of those fixtures across both _native_reader_confs and _java_reader_confs in test_parquet_testing_valid_files.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[greptile-apps]

Greptile Summary

Re-enables CPU/GPU parity coverage for five Parquet V2 DELTA-encoding fixtures now supported by cuDF, while retaining an expected failure for DELTA_LENGTH_BYTE_ARRAY on Spark versions before 3.4.

• Removes the five obsolete cuDF-linked expected-failure entries.
• Adds a Spark-version guard for the CPU reader’s DELTA_LENGTH_BYTE_ARRAY support boundary.
• Continues exercising each recovered fixture with native and Java footer-reader configurations.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

Important Files Changed

Filename	Overview
integration_tests/src/main/python/parquet_testing_test.py	Narrows Parquet fixture xfails by removing resolved cuDF limitations and retaining the Spark-version-specific CPU-reader exemption.

_{Reviews (5): Last reviewed commit: "Merge branch 'main' into codex/skip-reco..." | Re-trigger Greptile}

[wjxiz1992]

build

[thirtiseven]

LGTM

[wjxiz1992]

build