dart_lmdb2

A high-performance, embedded database for Dart applications, wrapping LMDB (Lightning Memory-Mapped Database). This package provides both high-level convenience methods and granular transaction control.

Linux	Windows	Android	MacOS	iOS	web
💙	💙	💙	💙	💙	-

Note: Precompiled native binaries for all platforms are bundled inside the Dart package, but for iOS/Android you need Flutter to run them on your mobile device. See flutter_lmdb2

Why LMDB?

LMDB is particularly well-suited for mobile and embedded applications because:

Mobile-Friendly Design:
- Database format is portable across all platforms
- Battery-efficient through memory mapping and minimal I/O operations
- No background processes or services required
- Predictable memory usage through configurable mapping size
- Instant app startup with lazy data loading
Minimal Resource Requirements:
- Ultra-compact native library (<150KB)
- Configurable DB ceiling with fixed mapSize limit
- Single-threaded design with minimal process overhead
- No additional runtime dependencies
Exceptional Performance:
- Zero-copy reads through direct memory mapping
- Efficient OS-level page caching eliminates I/O bottlenecks
- Optimized for read operations through direct memory access
- Read transactions never block writers
- Writers never block readers
- Sequential reads are extremely fast due to B+ tree design
Reliability:
- Full ACID compliance with atomic, crash-resistant transactions
- Copy-on-write design ensures database integrity even after crashes
- Direct page updates without separate log files
- Optional self-contained single file design simplifies backup operations
- Battle-tested in OpenLDAP

Why not LMDB ?

LMDB is not a general purpose database. It's a DB for specific purposes and shines in these areas.

You should not use LMDB, if

your schema is not K/V based, but relational
you need to efficiently query values additionally to keys
you want to save time series or streaming data (e.g. logging)
you cannot provide the same amount of RAM as your DB size in write scenarios

Scenarios:

Super fast querying of big DB's with minimal memory, e.g.
- Dictionaries
- Tile caches
- Texture caching
- Multiplatform assets
Coherent realtime data updates
- Producer -> Consumer(s)
  - Main App -> Plugin Instances
  - Audio Engine -> Visualizers
- Multiple Producers -> Multiple Consumers
  - Worker Pools
  - Distributed Processing
- Resource-Constrained IPC
  - Flutter App -> AudioUnit Extension
  - Host -> Sandboxed Plugins
Configuration data (and lots of it)

Supported Features

The following LMDB functionality is exposed:

Complete CRUD operations
Named databases for data organization
Cursor operations for range queries
Full transaction support with ACID guarantees
Comprehensive statistics and monitoring
Configurable initialization with all LMDB flags

LMDB Version

This package bundles LMDB version 0.9.70. While this version number hasn't changed in 3 years, LMDB is actively maintained. We track the exact git repository version: da9aeda.

Getting Started

Add the package to your pubspec.yaml:

dependencies:
  dart_lmdb2: ^0.9.5

Then run:

# For Dart projects:
dart pub get
# For Flutter projects:
flutter pub get

Usage

Import and use:

import 'package:lmdb2/lmdb2.dart';

void main() async {
    final db = ();
    await db.init('path/to/db');

    // Simple auto-transaction operations
    await db.putAuto('key', 'value'.codeUnits);
    final result = await db.getAuto('key');

    // Manual transaction control
    final txn = await db.txnStart();
    try {
        await db.put(txn, 'key1', 'value1'.codeUnits);
        await db.put(txn, 'key2', 'value2'.codeUnits);
        await db.putUtf8(txn, 'english_greeting', 'Hello World');
        await db.txnCommit(txn);
    } catch (e) {
        await db.txnAbort(txn);
        rethrow;
    }
}

Configure database settings using LMDBInitConfig:

final config = LMDBInitConfig(
  mapSize: 10 * 1024 * 1024,  // 10MB
  maxDbs: 1,
  envFlags: 0,
  mode: 0664,
);

await db.init('path/to/db', config: config);

Development

All native libraries are already bundled inside the directory lib/src/native:

> tree lib/src/native/
lib/src/native/
├── android
│   ├── arm64-v8a
│   │   └── liblmdb.so
│   └── x86_64
│       └── liblmdb.so
├── ios
│   └── liblmdb.a
├── linux
│   └── liblmdb.so
├── macos
│   └── liblmdb.dylib
└── windows
    └── lmdb.dll

You can rebuild these via the following steps:

Prerequisites

This package bundles LMDB to be built from source. The build process is automated but requires:

CMake (3.10 or higher)
C compiler (gcc, clang, or MSVC)
Dart SDK (3.0 or higher)

Platform-specific setup

Android

You need to have installed the Android NDK, ideally via Android Studio. Furthermore. you need to have the environment variable ANDROID_NDK_HOME set to the appropriate NDK location.

iOS / iPadOS / MacOS

Please install XCode and the XCode command line tools.

# Install build tools
brew install cmake

Linux

# Install build tools
sudo apt-get install build-essential cmake

Windows

Install Visual Studio with C++ development tools
Install CMake

Build & Test

Clone the repository and change to the project subdirectory:

git clone https://github.com/grammatek/dart_lmdb2.git
cd dart_lmdb2/dart_lmdb2

Install dependencies:

dart pub get

(Optionally) Rebuild the generated bindings via ffigen

dart run ffigen

Build the native library:

dart run tool/build.dart

For Android, you need to pass the flag --android:

dart run tool/build.dart --android

For iOS/iPadOS, you need to pass the flag --ios:

dart run tool/build.dart --ios

Run tests for active platform:

dart test

CREDITS

Many thanks to the OpenLDAP team to provide such a fantastic lightweight, portable and easy to use database. You can access the original source-code either directly as a GitHub standalone version or via the OpenLDAP GitLab repository.

LICENSE

MIT License - see LICENSE file