[Gabriel39]

What problem does this PR solve?

Related PRs: lance-format/lance-c#97, lance-format/lance-c#96

Update the pinned lance-c revision from 8e9e09ecc795c91ce7f97513db94c2d91eec799d to 4fafcd64525b274b113208390bbcca86bbbc093b, the latest upstream main commit when this PR was prepared, and update the archive checksum.

The new revision includes the upstream fix for cold-session vector prewarm when an index name matches a different schema column. It resolves the indexed field from segment metadata while preserving concurrent segment opening/prewarming. The revision also includes the row-address allowlist/blocklist prefilter APIs from lance-c #96.

Refresh one C++ test context in the existing Foyer patch to accommodate the test added by #96. The patch's added/deleted code lines are unchanged; no local prewarm patch is introduced.

Release note

Fix Lance index prewarm failures when a vector index name matches a different schema column.

Check List (For Author)

• Test
  • Manual test
    • Downloaded the exact pinned archive and verified its checksum through the Doris downloader.
    • bash thirdparty/test/lance-prefilter-patch-test.sh thirdparty/src: passed all scenarios, including fresh extraction, idempotence, cache invalidation, and patch-failure handling.
    • cargo metadata --locked --offline --format-version 1 on the patched source: passed dependency resolution, including Foyer.
    • Compiled the patched upstream C and C++ test sources with -fsyntax-only: passed.
    • bash -n on the dependency scripts and git diff --check: passed.
    • Full Doris build and runtime regression tests were not run locally; requested through run buildall.
• Behavior changed:
  • Yes. Includes the upstream prewarm fix and new prefilter APIs.
• Does this need documentation?
  • No.

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]

Reviewed the complete two-file diff at eeb54b2 and the exact imported lance-c changes (#96 and #97). No actionable introduced defect was found. There are no existing inline findings to carry forward and no additional user focus points.

Critical checkpoint conclusions:

• Purpose and evidence: the immutable pin includes the cold-session vector index/column-name collision fix and row allow/block APIs. Upstream regression sources cover the collision, shared-session reuse, snapshot behavior and row-mask correctness.
• Scope: the change is focused on the dependency URL/name/source/checksum and Foyer patch context. The overlay's added/deleted code is unchanged.
• Concurrency: the dataset lock is released before I/O; per-segment open and prewarm remain concurrent with fail-fast error propagation/cancellation. No new lock order or Doris thread entry is introduced.
• Lifetimes/static initialization: Arc-owned dataset snapshots, sessions and scanner masks retain their dependencies; no new cross-translation-unit initialization dependency was found.
• Configuration: no configuration item or dynamic-update behavior changes.
• ABI/storage compatibility: the C/C++ APIs are additive, opaque handles preserve the boundary, and the underlying Lance SDK revision and persisted formats are unchanged.
• Thrift/protobuf: no gensrc definitions change.
• Parallel paths: ordinary/colliding index names, vector/scalar/FTS prewarm, stable/nonstable row IDs, indexed/flat search, and synchronous/asynchronous/poll scanning were traced. Match and Phrase plan rewrites both retain the external mask.
• Conditional behavior: invalid inputs and scalar-segment/mask conflicts fail explicitly; same-field, missing-index and scalar/FTS cases preserve SDK handling.
• Test coverage: imported sources cover cold and reopened sessions, multiple segments, deterministic scheduling/cancellation, empty/replaced/combined masks, SQL filters and prepared FTS scoring. Additional pinned-SDK tests cover indexed vector filtering and mask execution; no build or runtime test was executed for this review.
• Expected results: no Doris regression-result files change; inspected assertions check selected rows, scores, errors, cache reuse and cancellation.
• Observability: existing FFI error reporting and underlying SDK diagnostics remain; no new Doris service path requires metrics. The colliding-vector workaround bypasses aggregate SDK prewarm info logs, without a substantiated correctness impact.
• Transactions/persistence: no EditLog, transaction commit/replay or persisted-metadata format changes.
• Writes/crash behavior: the imported changes operate on read/search/prewarm paths; no new data-write atomicity or crash-recovery mechanism is introduced.
• FE/BE variables: no new cross-process fields or propagation paths.
• Performance: concurrent segment I/O is retained; row masks filter before ranking and survive prepared-plan reconstruction. No actionable new performance regression was established.
• Other integration checks: the downloaded archive matches MD5 65e924fb62137c5dd4d58af3fb33923a; all 15 Foyer patch targets pass a zero-fuzz dry run. The unchanged upstream manifests/lockfile, in-memory patched lock graph, downloader invalidation and third-party-only BE linkage were inspected.

Validation limit: this is a static review, including archive verification and patch dry-run only. Doris/upstream builds, link checks, shell test suites and runtime regressions were not run. No unresolved candidate or uncovered static-review scope remains.

Review converged after 1 round(s).

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 100% (0/0) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	77.20% (35840/46426)
Line Coverage	62.47% (404498/647543)
Region Coverage	58.92% (341502/579564)
Branch Coverage	59.90% (157975/263750)