r2d2-sqlite

r2d2-sqlite is a SQLite connection-pool adapter for the r2d2 generic connection pool in Rust. It implements r2d2’s ConnectionManager for rusqlite, so you can maintain a pool of reusable SQLite connections - to a file, an in-memory database, or a shared-cache database - and hand them out to worker threads efficiently. Modeled on Steven Fackler’s r2d2-postgres, it removes the cost of repeatedly opening and configuring SQLite connections in concurrent applications.

What You Get

• A SqliteConnectionManager implementing r2d2’s ConnectionManager for rusqlite
• Constructors for file-based, in-memory, and named shared-cache SQLite databases
• Hooks to set open flags and run per-connection initialization (for example, PRAGMAs)
• Reusable pooled connections that avoid repeated open and setup overhead

Common Use Cases

• Serving concurrent web requests that each need a SQLite connection
• Sharing an in-memory SQLite database across multiple worker threads
• Applying consistent PRAGMA settings to every connection in an application

Under The Hood

Architecture

The crate is a thin adapter: a single SqliteConnectionManager struct captures a connection source (a file path, plain in-memory, or a named shared-cache database) together with optional OpenFlags and an initialization closure. It implements r2d2’s ConnectionManager trait, whose connect method opens a fresh rusqlite Connection, applies the flags, and runs the init hook, while is_valid and has_broken let the pool recycle connections. r2d2 itself owns the pooling, checkout, and lifecycle logic.

Tech Stack

Written in Rust on top of rusqlite for the SQLite bindings and r2d2 for the generic pool, with uuid used to generate unique names for shared-cache in-memory databases. It is a compact library crate targeting stable Rust, with an update-docs.sh helper and a changelog tracking releases against rusqlite versions.

Code Quality

The implementation is a single src/lib.rs file kept intentionally small, with a tests/test.rs suite covering file, memory, and shared-cache pooling paths. Because it delegates the hard concurrency work to r2d2 and the SQL work to rusqlite, its own surface is easy to audit; the main maintenance signal is keeping the rusqlite version dependency current, which the changelog shows happening.

API Design

The API is minimal and familiar to anyone who has used r2d2: build a SqliteConnectionManager, pass it to r2d2::Pool::new, and check out rusqlite connections. Helper constructors (file, memory) and the with_flags/with_init builders keep configuration readable, so getting a working pool takes only a few lines.