Laravel Tarantool Laravel Package

Product Decisions This Supports

• Backend selection for high-throughput paths: This package is the Laravel-side integration for Tarantool (in-memory / tuple-store). It supports the decision to place latency-sensitive features — session storage, rate limiters, counters, leaderboards, real-time feeds — behind Tarantool while keeping Eloquent/Query Builder ergonomics in application code.
• Build vs. buy on the data-access layer: The 1.1.2 patch (wrapUnion overridden to strip parentheses around UNION subqueries) is a concrete data point for the buy-side case. Tarantool's SQL dialect diverges from standard SQL, and someone has to own that divergence. Buying means a third party absorbs dialect quirks; building means your team owns a forkable adapter plus its test matrix. Note the patch is small and surgical, which is evidence the maintainer is responsive but also that the abstraction is thin — expect to hit dialect edges yourself.
• Portability debt on the roadmap: Any feature that relies on UNION-inside-subquery construction now behaves differently under this driver than under MySQL/Postgres. If your roadmap includes multi-store support or engine migration, budget for a query-compatibility layer or a test suite that runs against both drivers.
• Release/upgrade policy decisions: 1.1.2 is a patch-level, non-breaking correctness fix with no API surface change — it should pass a standard dependency-bump gate. It also argues for pinning the driver version in composer.json and covering UNION queries with integration tests, since SQL-generation behavior is now driver-specific.
• Prioritization of test coverage: The change only matters if you emit UNIONs in subqueries. If you do, this is a bug fix that unblocks a feature; if you don't, it is inert. That distinction should drive whether it is a "ship now" or "ride the next routine bump."

When to Consider This Package

Adopt when:

• Tarantool is already (or is being deliberately) part of the architecture and you want Laravel-native models, query building, and connection management rather than hand-rolling a client wrapper.
• Your team is small and the alternative is owning a bespoke connector plus its maintenance and dialect edge cases.
• Your query patterns are conventional (CRUD, filters, pagination) and you can tolerate the driver's SQL-generation differences from core Laravel.
• You need a maintained, versioned dependency that tracks Laravel releases; the cadence of small patch releases like 1.1.2 is a reasonable health signal.

Look elsewhere when:

• You need one query layer that behaves identically across MySQL/Postgres and Tarantool — a lowest-common-denominator ORM or an explicit repository boundary is a better fit than a driver that silently overrides SQL generation.
• You depend heavily on advanced UNION semantics, complex subquery composition, or other constructs where this release shows the dialect will diverge. Every such gap becomes your bug.
• You require vendor-backed SLA support, or your organization cannot carry a niche-ecosystem dependency in a critical path.
• Tarantool is not in the stack — there is no scenario where adopting this package is a net positive without the underlying store.
• Your workload needs features Tarantool does not model well (ad-hoc analytical joins, heavy relational integrity), regardless of driver quality.

Specific to 1.1.2: treat it as a safe patch upgrade. Verify any code path that builds UNIONs in subqueries, and avoid copy-pasting generated SQL between this driver and a standard SQL driver in docs or runbooks — the parenthesis behavior differs by design.

How to Pitch It (Stakeholders)

For executives: This is a small, low-risk maintenance release (1.1.2) for our Laravel-to-Tarantool data-access package. It fixes a correctness issue where UNION subqueries were generated with parentheses that Tarantool rejects. There are no API or breaking changes, no migration work, and no user-visible behavior change outside that specific query shape. The broader message is that this dependency is actively maintained but sits in a narrow ecosystem — it is the right call only while Tarantool remains part of our architecture, and it carries a modest portability cost we should keep visible on the roadmap.

For engineering: Patch-level bump, semver-safe. The change overrides the framework's default wrapUnion to drop parentheses around union subqueries, matching Tarantool's SQL grammar. Action items: add or confirm an integration test covering UNION-in-subquery queries, pin the driver version, and document that generated SQL from this driver is not byte-identical to core Laravel's. Flag any shared query builders or raw-SQL snippets that assume standard parenthesis behavior.