Mutating a Substring apparently makes a copy of the entire original String (in some scenarios)

substr-mutation.swift

// Make sure both the string and the substring are larger than 15 UTF-8 bytes
// to avoid the small string optimization
var str = "Hello world 1 Hello world 2 Hello world 3 Hello world 4 Hello world 5 Hello world 6 Hello world 7 Hello world 8 Hello world 9"
let prefixToStrip = 14
var substr = str.dropFirst(prefixToStrip).prefix(27)
let strToAppend = "+++APPENDED+++"

// ⚠️ It makes a difference how you mutate the Substring:
// - substr.append → Apparently makes a copy of the entire original string
//   *and* even shifts the original string contents back to make room,
//   i.e. making append an O(n) operation where `n == str.count` (bad).
// - substr.append(contentsOf:) → Apparently only copies the substring part (good)
for char in strToAppend {
    substr.append(char)
}
//substr.append(contentsOf: strToAppend)

print("Substring")
print("---------")
print("contents:", substr)
print("utf8.count:", substr.utf8.count)
print("isContiguousUTF8:", substr.isContiguousUTF8)
substr.withUTF8 { utf8Buffer in
    print("utf8Buffer.baseAddress:", utf8Buffer.baseAddress!)
    // Unsafe!!! Assume that the Substring copied the entire original collection
    // on mutation, so we can get back the full buffer by extending the pointer
    // out of bounds.
    let fullBuffer = UnsafeBufferPointer(
        start: utf8Buffer.baseAddress! - prefixToStrip,
        count: str.utf8.count + strToAppend.utf8.count
    )
    let newString = String(decoding: fullBuffer, as: UTF8.self)
    print(newString)
}

print("")
print("Original string")
print("---------")
print("isContiguousUTF8:", str.isContiguousUTF8)
str.withUTF8 { utf8Buffer in
    print("utf8Buffer.baseAddress:", utf8Buffer.baseAddress!)
}

Typical output for me with the str.append(char) loop (note that thefullBuffer is printed correctly, indicating that the "real" buffer is larger than the segment we can access through the mutated Substring):

Substring
---------
contents: Hello world 2 Hello world 3+++APPENDED+++
utf8.count: 41
isContiguousUTF8: true
utf8Buffer.baseAddress: 0x000000013df0970e
Hello world 1 Hello world 2 Hello world 3+++APPENDED+++ Hello world 4 Hello world 5 Hello world 6 Hello world 7 Hello world 8 Hello world 9

Original string
---------
isContiguousUTF8: true
utf8Buffer.baseAddress: 0x00000001092bc100

Typical output for me with str.append(contentsOf:) (note that the full buffer prints garbage, indicating, that we're going past the bounds of the buffer):

Substring
---------
contents: Hello world 2 Hello world 3+++APPENDED+++
utf8.count: 41
isContiguousUTF8: true
utf8Buffer.baseAddress: 0x0000600003fe0500
������)�������Hello world 2 Hello world 3+++APPENDED+++�����������������������@���<��������������������������������������������������������

Original string
---------
isContiguousUTF8: true
utf8Buffer.baseAddress: 0x0000000107ecc100