以不同方式创建数组时,反初始化顺序不同
Deinitialization order is different when creating array differently
TL;DR: 为什么,创建数组时连同初始值一起去初始化,直接去初始化,但是创建完数组后填充数组值产生不同的行为?
我在JavaScript之后学习Swift,它有GC,所以ARC和反初始化的概念对我来说有点新手。为了更深入地理解它,我根据 example from the official documentation:
编写了以下代码
// definitions; see the next two snippets for usage
class Bank {
static var coins: Int = 100 {
didSet {
let diff = coins - oldValue
let sign = diff < 0 ? "-" : "+"
print("Bank: \(oldValue) \(sign) \(diff.magnitude) = \(coins)")
}
}
static func give(_ amount: Int, to user: User) {
guard amount <= coins else {
print("[ERROR] The bank does not have enough coins; requested \(amount - coins) more coins than available")
return
}
user.purse += amount
coins -= amount
}
static func take(_ amount: Int, from user: User) {
user.purse -= amount
coins += amount
}
}
class User {
private static var count = 0
let id: Int
var purse = 0
init() {
User.count += 1
id = User.count
}
deinit {
print("User #\(id) perished")
Bank.take(purse, from: self)
}
}
当我使用现有用户实例 (array = [value, value]
) 创建数组时,取消初始化发生在分配给 nil
(array[0] = nil
):
之后
var users: [User?] = [
User(),
User(),
]
Bank.give(90, to: users[0]!)
users[0] = nil
Bank.give(50, to: users[1]!) // works
// Bank: 100 - 90 = 10
// User #1 perished
// Bank: 10 + 90 = 100
// Bank: 100 - 50 = 50
… 但是当我首先创建一个空数组然后用值填充它时 (array = []; array[0] = value; array[1] = value
),取消初始化会在稍后的某个随机点发生:
var users = [User?](repeating: nil, count: 2)
users[0] = User()
users[1] = User()
Bank.give(90, to: users[0]!)
users[0] = nil
Bank.give(50, to: users[1]!) // fails
// Bank: 100 - 90 = 10
// [ERROR] The bank does not have enough coins; requested 40 more coins than available
// User #1 perished
// Bank: 10 + 90 = 100
从编译器的角度来看有什么区别?造成这种差异的原因是什么?
TL;DR 不要依赖 deallocation/de-initialization 时机。
确切地弄清楚何时释放,因此 de-initialization 发生在之前,发生是一个 non-trivial 过程。
线程(几乎每个应用程序都是 multi-threaded,即使它不直接使用线程)、临时编译器、传递值等都有助于准确混淆最后一次引用对象的时间消失。
除此之外,Swift 游乐场本身可能会保留 non-obvious 对象的引用,因为它的工作方式,这似乎是这里发生的事情。将您的代码放入 Swift 命令行应用程序对我们有用,但是 YMMV!
HTH
TL;DR: 为什么,创建数组时连同初始值一起去初始化,直接去初始化,但是创建完数组后填充数组值产生不同的行为?
我在JavaScript之后学习Swift,它有GC,所以ARC和反初始化的概念对我来说有点新手。为了更深入地理解它,我根据 example from the official documentation:
编写了以下代码// definitions; see the next two snippets for usage
class Bank {
static var coins: Int = 100 {
didSet {
let diff = coins - oldValue
let sign = diff < 0 ? "-" : "+"
print("Bank: \(oldValue) \(sign) \(diff.magnitude) = \(coins)")
}
}
static func give(_ amount: Int, to user: User) {
guard amount <= coins else {
print("[ERROR] The bank does not have enough coins; requested \(amount - coins) more coins than available")
return
}
user.purse += amount
coins -= amount
}
static func take(_ amount: Int, from user: User) {
user.purse -= amount
coins += amount
}
}
class User {
private static var count = 0
let id: Int
var purse = 0
init() {
User.count += 1
id = User.count
}
deinit {
print("User #\(id) perished")
Bank.take(purse, from: self)
}
}
当我使用现有用户实例 (array = [value, value]
) 创建数组时,取消初始化发生在分配给 nil
(array[0] = nil
):
var users: [User?] = [
User(),
User(),
]
Bank.give(90, to: users[0]!)
users[0] = nil
Bank.give(50, to: users[1]!) // works
// Bank: 100 - 90 = 10
// User #1 perished
// Bank: 10 + 90 = 100
// Bank: 100 - 50 = 50
… 但是当我首先创建一个空数组然后用值填充它时 (array = []; array[0] = value; array[1] = value
),取消初始化会在稍后的某个随机点发生:
var users = [User?](repeating: nil, count: 2)
users[0] = User()
users[1] = User()
Bank.give(90, to: users[0]!)
users[0] = nil
Bank.give(50, to: users[1]!) // fails
// Bank: 100 - 90 = 10
// [ERROR] The bank does not have enough coins; requested 40 more coins than available
// User #1 perished
// Bank: 10 + 90 = 100
从编译器的角度来看有什么区别?造成这种差异的原因是什么?
TL;DR 不要依赖 deallocation/de-initialization 时机。
确切地弄清楚何时释放,因此 de-initialization 发生在之前,发生是一个 non-trivial 过程。
线程(几乎每个应用程序都是 multi-threaded,即使它不直接使用线程)、临时编译器、传递值等都有助于准确混淆最后一次引用对象的时间消失。
除此之外,Swift 游乐场本身可能会保留 non-obvious 对象的引用,因为它的工作方式,这似乎是这里发生的事情。将您的代码放入 Swift 命令行应用程序对我们有用,但是 YMMV!
HTH