Cover trie index modes

[zaidoon1]

Part 12 of 13 in the UDI split.

Stack order:

1. Add IndexFactory compatibility names #14954 Add IndexFactory compatibility names
2. Add UDI index mode API vocabulary #14959 Add UDI index mode API vocabulary
3. Add optional UDI builder protocols #14960 Add optional UDI builder protocols
4. Add built-in index factory wrappers #14961 Add built-in index factory wrappers
5. Add built-in index factory tests #14969 Add built-in index factory tests
6. Promote IndexFactory as UDI SPI #14962 Promote IndexFactory as UDI SPI
7. Wire IndexFactory through block-based tables #14963 Wire IndexFactory through block-based tables
8. Add IndexFactory table routing tests #14964 Add IndexFactory table routing tests
9. Add IndexFactory parallel routing tests #14970 Add IndexFactory parallel routing tests
10. Account UDI blocks as index blocks #14965 Account UDI blocks as index blocks
11. Support parallel AddIndexEntry in trie index #14966 Support parallel AddIndexEntry in trie index
12. Cover trie index modes #14967 Cover trie index modes
13. Add IndexFactory stress and benchmark flags #14968 Add IndexFactory stress and benchmark flags

Previous: #14966.
Next: #14968.

Depends on #14966. Because this PR targets facebook/rocksdb:main, GitHub shows a cumulative diff until earlier PRs land. After those land, the intended review diff is:

zaidoon/udi-09-trie-parallel-builder..zaidoon/udi-10-trie-index-mode-tests

That final review delta is 1 file with 707 insertions and 280 deletions.

What changed:

• Add DB-level coverage for trie index mode combinations.
• Keep trie mode test coverage separate from the trie builder implementation.

Validation:

• AUTO_CLEAN=1 make -j14 trie_index_db_test
• ./trie_index_db_test
• make check-sources

[github-actions]

✅ clang-tidy: No findings on changed lines

Completed in 1094.9s.

[github-actions]

✅ Claude Code Review

Auto-triggered after CI passed — reviewing commit cbc3131

---

Summary

This is Part 12 of 13 in the UDI (User Defined Index) split. The cumulative diff is large (50 files, ~6000 insertions), but the PR-specific delta is 1 file (trie_index_db_test.cc) with ~707 insertions and ~280 deletions. The cumulative diff introduces the IndexFactory API refactoring (renaming UserDefinedIndexFactory to IndexFactory, adding IndexMode enum, ReadIndex enum, BuiltinIndexFactoryBuilder, parallel compression support for custom indexes, and standard_index_is_stub table property). The test file adds comprehensive DB-level coverage for trie index mode combinations. Overall this is a well-structured change with thorough test coverage.

High-severity findings (0):

No high-severity findings.

Full review (click to expand)

Findings

🔴 HIGH

None.

🟡 MEDIUM

M1. C API SyncIndexModeFromLegacyUdiOptions priority inversion -- db/c.cc:468

The SyncIndexModeFromLegacyUdiOptions() method checks use_udi_as_primary_index before fail_if_no_udi_on_open. If a caller sets both flags true then clears use_udi_as_primary_index, index_mode becomes kStandardRequired because fail_if_no_udi_on_open is still true. This is correct behavior but the C API test at db/c_test.c:3423 only tests each flag independently.

• Suggested fix: Add a C API test that sets both flags, then clears one, to verify the interaction.

M2. Test parameterization covers only secondary/primary but not kCustomOnly or kStandardRequired -- trie_index_db_test.cc:4727

The test is parameterized with ::testing::Bool() (secondary vs primary mode), corresponding to kStandardDefault and kCustomDefault. The new IndexMode enum adds kCustomOnly and kStandardRequired, but these are not exercised by the DB-level tests despite the PR title "Cover trie index modes."

• Suggested fix: Add parameterization or dedicated test cases for kCustomOnly and kStandardRequired modes, or add a TODO comment if planned for Part 13.

M3. OpenDBImpl and TrieIndexReadOptions() still use legacy APIs -- trie_index_db_test.cc:188,206

The test fixture uses table_options.use_udi_as_primary_index and ro.table_index_factory rather than the new IndexMode/ReadIndex enums directly. This tests the legacy->enum translation but not the new API paths. The db_wide_blob_direct_write_test.cc changes show migration to read_index = kPreferCustom, but the main test file doesn't follow suit.

• Suggested fix: Add test paths using the new API directly (index_mode = kCustomDefault, read_index = kPreferCustom).

🟢 LOW / NIT

L1. CF handle cleanup uses delete instead of DestroyColumnFamilyHandle -- trie_index_db_test.cc:4583

In MultiCFCoalescingIterator, delete cf1; delete cf2; should be db_->DestroyColumnFamilyHandle(cf1/cf2) to match the pattern in MultipleColumnFamilies at line 1999.

L2. IndexMode enum value ordering -- include/rocksdb/table.h

kStandardRequired = 4 placed after kCustomOnly = 3 with a backward-compatibility comment. Logically kStandardRequired sits between kStandardDefault and kCustomDefault in the migration path but numerically comes last, which could confuse users comparing enum values.

Positive Observations

• Comprehensive test coverage across flush, compaction, WAL replay, multi-CF, ingestion, prefix iteration, direction switching, same-user-key blocks, DeleteRange, MultiGet, coalescing iterators, and edge cases
• Dual-index verification pattern catches divergence between standard and trie paths
• Stress-like patterns (BatchedPrefixScanStressLike) mimic crash-test workloads
• Clean C API bridge via SyncIndexModeFromLegacyUdiOptions()
• All three build systems (Make, CMake, BUCK) updated consistently

---

ℹ️ About this response

Generated by Claude Code.
Review methodology: claude_md/code_review.md

Limitations:

• Claude may miss context from files not in the diff
• Large PRs may be truncated
• Always apply human judgment to AI suggestions

Commands:

• /claude-review [context] — Request a code review
• /claude-query <question> — Ask about the PR or codebase