fix!: bump IVF_RQ version for compatibility check

[BubbleCal]

Summary

• add IVF_RQ compatibility tests that run against the latest stable pylance release to ensure read/query compatibility both directions
• update IndexDetails to report the actual highest vector version rather than a single constant

[github-actions]

ACTION NEEDED
Lance follows the Conventional Commits specification for release automation.

The PR title and description are used as the merge commit message. Please update your PR title and description to match the specification.

For details on the error please inspect the "PR Title Check" action.

[github-actions]

Review

P1: index_version derived from all old_indices, not just the merged subset

In rust/lance/src/index/append.rs, the new code takes old_indices.iter().map(|idx| idx.index_version).max() over all old indices, but only the last indices_merged entries are being merged. The fragment bitmap correctly scopes to merged_indices, but the version derivation doesn't. While in practice all entries in old_indices should be the same type/version (validated at line 102), using merged_indices.iter() instead of old_indices.iter() would be more precise and consistent with the bitmap logic:

let index_version = merged_indices
    .iter()
    .map(|idx| idx.index_version)
    .max()
    .ok_or_else(|| ...)?;

P1: max_vector_version() manually enumerates all vector types

IndexType::max_vector_version() lists all vector index variants by hand. If a new vector index type is added without updating this list, the max version will be silently wrong. Consider either:

• Deriving the list from IndexType variants (e.g., using strum or a similar enum iterator), or
• Adding a compile-time or test assertion that covers all IndexType variants.

Nit: Compat test missing check_write

IvfRqVectorIndex only has check_read — no check_write (unlike the IvfPqVectorIndex test above it). If the intent is to also verify write-path compatibility (append + optimize on an index created by a different version), consider adding it.

[Xuanwo]

Hi @BubbleCal, I think the review raised by claude is valid:

index_version derived from all old_indices, not just the merged subset

Do we need to take care about this?

[codecov]

Codecov Report

❌ Patch coverage is 91.42857% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
rust/lance/src/index.rs	25.00%	2 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

[BubbleCal]

Hi @BubbleCal, I think the review raised by claude is valid:

index_version derived from all old_indices, not just the merged subset

Do we need to take care about this?

it's common that we just append a delta index and merged_num=0 in that case, so we need to derive index_version from all old_indices.

I think today all old_indices will always be with the same version, even on a new version of lance, we always respect the existing index's version when incremental indexing

[Xuanwo]

Thank you for this PR!

[wjones127]

Okay this should go good to go. Will merge once CI is green and we can make a release.