avnerbarr/FlockKit

Features

• ✅ Simple, Swifty API over flock(2)
• ✅ Exclusive (write) and shared (read) locks
• ✅ Blocking and non-blocking acquisition
• ✅ Convenience helpers for “run this while locked”
• ✅ Linux + Darwin (macOS, iOS, tvOS, watchOS) friendly
• ⚠️ Designed for cross-process coordination, not intra-process thread safety

Requirements

• Swift: Swift 5.9 or newer (adjust as appropriate)
• Platforms: Any platform that exposes POSIX flock(2) via:

- Darwin (macOS, iOS, tvOS, watchOS) - Glibc (Linux)

Installation

Swift Package Manager

Add FlockKit as a dependency in your Package.swift:

// swift-tools-version: 5.9

import PackageDescription
let package = Package(
name: "YourApp",
dependencies: [
    .package(
        url: "https://github.com/avnerbarr/FlockKit.git",
        from: "0.1.0"
    )
],
targets: [
    .target(
        name: "YourApp",
        dependencies: [
            .product(name: "FlockKit", package: "FlockKit")
        ]
    )
]

Xcode

1. File -> Add Packages...
2. Enter the URL: https://github.com/avnerbarr/FlockKit.git
3. Select the latest version and add it to your target.

Quick start

Basic exclusive lock (blocking)

import FlockKit

do {
    let lock = try FileLock(path: "/tmp/myapp.lock")
    try lock.lockExclusive() // blocks until the lock is available
    defer { lock.unlock() }

    // Only one *process* at a time will reach this point.
    print("Running critical section…")
    // do important work
} catch {
    print("Failed to acquire lock: \(error)")
}

Non-blocking lock

Use blocking: false to “try once and bail” instead of waiting:

import FlockKit

do {
    let lock = try FileLock(path: "/tmp/myapp.lock")
    try lock.lockExclusive(blocking: false)
    defer { lock.unlock() }
    print("We own the lock, running job…")
} catch let error as FlockKitError {
    switch error {
    case .lockFailed:
        print("Another instance is already running, exiting.")
    default:
        print("Lock error: \(error)")
    }
}

Scoped helpers

If you just want to “run this closure while locked”, use the helper functions:

Exclusive (write) lock around a job

import FlockKit

do {
    let result = try withExclusiveWriteLock(at: "/tmp/myapp.job.lock") {
        try runJob()
    }
} catch let error as LockedJobError {
    switch error {
    case .flock(let lockError):
        print("Failed to acquire lock: \(lockError)")
    case .job(let jobError):
        print("Job failed while holding lock: \(jobError)")
    }
}

Shared (read) lock

import FlockKit

do {
    let result = try withSharedLock(at: "/tmp/myapp.cache.lock") {
        try loadCache()
    }
} catch let error as LockedJobError {
    switch error {
    case .flock(let lockError):
        print("Failed to acquire lock: \(lockError)")
    case .job(let jobError):
        print("Job failed while holding lock: \(jobError)")
    }
}

Both helpers:

• Create a new FileLock for the given path,
• Acquire the requested lock (exclusive or shared),
• Run your closure,
• Always call unlock() in a defer block,
• Wrap outcomes in LockedJobError so you know whether locking or the job failed.

Semantics & Caveats

FlockKit is for cross-process mutual exclusion. not per-thread synchronization

Locks are tracked by the OS per process, not per thread.

• If two different processes each create a FileLock for the same path and try to take an exclusive lock, only one will succeed at a time.
• If two threads in the same process each create their own FileLock for the same path and call lockExclusive(), both calls can succeed: the process already owns the lock, so the second flock() is effectively a no-op.

It is not a general replacement for a mutex between threads in the same process.

For thread-level synchronisation inside a single process, use standard Swift concurrency primitives:

• DispatchQueue
• NSLock
• os_unfair_lock
• pthread_mutex_t
• Swift concurrency (actors, etc.)

You can combine those with FileLock if you need both in-process and cross-process coordination.

FileLock is not thread-safe

The FileLock type keeps internal mutable state (isLocked) without any synchronisation. If you call its methods concurrently from multiple threads on the same instance, the behaviour is undefined and the internal bookkeeping may disagree with the OS lock.

If you must share a FileLock instance between threads, wrap all access in your own synchronisation (e.g. a serial DispatchQueue or NSLock).

Advisory locks

flock(2) locks are advisory:

• Other processes must also use flock() (or FlockKit) on the same path for the lock to have effect. But you probably wrote those applications, so it shouldn't be a problem
• A process that ignores the lock and writes directly to the file is not prevented by the kernel from doing so.

Example: single-instance CLI tool

A simple pattern for “only one instance may run at a time”:

import FlockKit

@main
struct MyTool {
    static func main() {
        do {
            try withExclusiveWriteLock(at: "/tmp/mytool.run.lock", blocking: false) {
                try runCommand()
            }
        } catch let error as LockedJobError {
            switch error {
            case .flock:
                fputs("Another instance is already running.\n", stderr)
                exit(1)
            case .job(let jobError):
                fputs("Command failed: \(jobError)\n", stderr)
                exit(2)
            }
        } catch {
            fputs("Unexpected error: \(error)\n", stderr)
            exit(3)
        }
    }
}

License

FlockKit is licensed under the MIT License.

Package Metadata

Repository: avnerbarr/FlockKit

Stars: 1

Forks: 0

Open issues: 0

Default branch: main

Primary language: swift

License: MIT

README: README.md