Crate dbkit 

Multi-backend database infrastructure.

Transactional writes go through sqlx’s Any driver, so the backend (Postgres, MySQL, or SQLite) is selected from the connection URL scheme. Analytical reads go through a pluggable Arrow-based engine (DuckDB or DataFusion), behind feature flags.

• ConnectionManager — sqlx connection pool with auto-create DB
• Cache — DashMap-based concurrent key-value cache with named buckets
• BaseHandler — execute_write (sqlx) / execute_read (Arrow) / execute_read_as (typed), plus transactional → analytical sync
• InitializationHandler — backend-aware DDL migration executor

Example

use dbkit::ConnectionManager;

// The URL scheme selects the backend: postgres:// , mysql:// , sqlite://
let conn = ConnectionManager::new("postgres://localhost/myapp").await?;

// `pool()` yields a sqlx `AnyPool`, usable across all supported backends.
let _pool = conn.pool();

Note: the write/read handlers and migration runner are being ported onto this pool; see the crate changelog for the current status.

Re-exports

pub use config::Backend;

pub use config::ConfigBuilder;

pub use config::DbkitConfig;

pub use config::SslMode;

Modules

config

Structs

AnyRow

BaseHandler

Core query executor: transactional writes via sqlx, and optionally analytical reads via a pluggable [ReadEngine] (DuckDB or DataFusion).

Cache

Concurrent key-value cache with named buckets.

ConnectionManager

Multi-backend connection pool with optional automatic database creation.

InitializationHandler

Batch DDL migration executor with tracking.

PoolStatus

Snapshot of connection pool health.

Enums

DbValue

A backend-neutral parameter value.

DbkitError

FetchMode

How many rows to expect from a query.

QueryResult

Result wrapper for write queries.

WriteOp

Unified write operation types.

Type Aliases

AnyPool