What are the differences between throws and rethrows in Swift? - swift

Question

After searching for some references to figure it out, -unfortunately- I could not find useful -and simple- description about understanding the differences between throws and rethrows. It is kind of confusing when try to understand how we should use them.
I would mention that I am kind of familiar with the -default- throws with its simplest form for propagating an error, as follows:
enum CustomError: Error {
case potato
case tomato
}
func throwCustomError(_ string: String) throws {
if string.lowercased().trimmingCharacters(in: .whitespaces) == "potato" {
throw CustomError.potato
}
if string.lowercased().trimmingCharacters(in: .whitespaces) == "tomato" {
throw CustomError.tomato
}
}
do {
try throwCustomError("potato")
} catch let error as CustomError {
switch error {
case .potato:
print("potatos catched") // potatos catched
case .tomato:
print("tomato catched")
}
}
So far so good, but the problem arises when:
func throwCustomError(function:(String) throws -> ()) throws {
try function("throws string")
}
func rethrowCustomError(function:(String) throws -> ()) rethrows {
try function("rethrows string")
}
rethrowCustomError { string in
print(string) // rethrows string
}
try throwCustomError { string in
print(string) // throws string
}
what I know so far is when calling a function that throws it has to be handled by a try, unlike the rethrows. So what?! What is logic that we should follow when deciding to use throws or rethrows?

Answer 1

From "Declarations" in the Swift book:
Rethrowing Functions and Methods
A function or method can be declared with the rethrows keyword to
indicate that it throws an error only if one of its function
parameters throws an error. These functions and methods are known as
rethrowing functions and rethrowing methods. Rethrowing functions and
methods must have at least one throwing function parameter.
A typical example is the map method:
public func map<T>(_ transform: (Element) throws -> T) rethrows -> [T]
If map is called with a non-throwing transform, it does not throw
an error itself and can be called without try:
// Example 1:
let a = [1, 2, 3]
func f1(n: Int) -> Int {
return n * n
}
let a1 = a.map(f1)
But if map is called with a throwing closure then itself can throw
and must be called with try:
// Example 2:
let a = [1, 2, 3]
enum CustomError: Error {
case illegalArgument
}
func f2(n: Int) throws -> Int {
guard n >= 0 else {
throw CustomError.illegalArgument
}
return n*n
}
do {
let a2 = try a.map(f2)
} catch {
// ...
}
If map were declared as throws instead of rethrows then you would
have to call it with try even in example 1,
which is "inconvenient" and bloats the code unnecessary.
If map were declared without throws/rethrows then you could not
call it with a throwing closure as in example 2.
The same is true for other methods from the Swift Standard Library
which take function parameters: filter(), index(where:), forEach() and many many more.
In your case,
func throwCustomError(function:(String) throws -> ()) throws
denotes a function which can throw an error, even if called with
a non-throwing argument, whereas
func rethrowCustomError(function:(String) throws -> ()) rethrows
denotes a function which throws an error only if called with a
throwing argument.
Roughly speaking, rethrows is for functions which do not throw
errors "on their own", but only "forward" errors from their function
parameters.

Answer 2

Just to add something along with Martin's answer. A non throwing function with the same signature as a throwing function is considered a sub-type of the throwing function. That is why rethrows can determine which one it is and only require try when the func param also throws, but still accepts the same function signature that doesn't throw. It's a convenient way to only have to use a do try block when the func param throws, but the other code in the function doesn't throw an error.