Swiftian Satire, or Tragedy?

I honestly cannot tell if Swift developers are seriously eating Irish babies, or taking the Mickey.

From bad implementations of Equatable and Hashable, the wag leaps to:

extension GridPoint : HashVisitable {
    func hash<H: Hasher>(_ hasher: inout H) {
        self.x.hash(&hasher)
        self.y.hash(&hasher)
    }
}

Bravo! That’s easily the funniest punchline to a programming joke since “where do you think the chaos came from?”

Starts with the most bizarre strawman Objective-C Sieve of Erathosthenes I’ve ever seen. A real implementation would be in C, because Obj-C is C with objects, and it’d be massively faster:

#include <stdlib.h>
#include <stdio.h>

typedef char BOOL; // or link in Foundation
#define YES 1
#define NO 0

int main(int argc, char **argv) {
    if (argc != 2) {
        printf("Usage: primes COUNT\n");
        exit(1);
    }
    long n = atoi(argv[1]);
    BOOL p[n];
    p[0] = NO;
    p[1] = NO;
    for (long i = 2; i < n; ++i) {
        p[i] = YES;
    }
    for (long i = 2; i < n; ++i) {
        for (long j = i*i; j < n; j += i) {
            p[j] = NO;
        }
    }
    for (long i = 1; i < n; ++i) {
        if (p[i]) {
            printf("%ld ", i);
        }
    }
    puts("");
    return 0;
}

This does use more lines of code, but they’re short, low-density, and it’s instantly obvious what it’s doing (my predilection for 1-char var names aside). Can you actually decode his filter-based version?

func sieve(_ sorted: [Int]) -> [Int] {
    guard !sorted.isEmpty else { return [] }
    let (head, tail) = (sorted[0], sorted[1..<sorted.count])
    return [head] + sieve(tail.filter { $0 % head > 0 })
}

let numbers = Array(2...1000000)
let primes = sieve(numbers)
print(primes)

And the runtime experiment:

mdh@Aegura:~/Code/CodeC% time clang -O3 -o primes primes.c                
clang -O3 -o primes primes.c  0.04s user 0.41s system 83% cpu 0.534 total
mdh@Aegura:~/Code/CodeC% time ./primes 1000000 >~/Desktop/primes.txt      
./primes 1000000 > ~/Desktop/primes.txt  0.02s user 0.00s system 89% cpu 0.025 total

mdh@Aegura:~/Code/CodeSwift% time swiftc -O -o swiftPrimes swiftPrimes.swift
swiftc -O -o swiftPrimes swiftPrimes.swift  0.57s user 0.64s system 69% cpu 1.754 total
mdh@Aegura:~/Code/CodeSwift% time ./swiftPrimes >~/Desktop/swiftPrimes.txt  
./swiftPrimes > ~/Desktop/swiftPrimes.txt  51.79s user 26.49s system 99% cpu 1:18.78 total

So the naïve C implementation is about 3,151x faster. I can’t measure it precisely because a limit measurable in C, would take Swift until the heat death of the Universe.

So here’s my question: Is Vincent aware of this, and his theme of “diabetes”, “sugar”, “saccharine”, etc. pointing at how fat, bloated, slow, and deadly Swift is? He never lets on if this is a joke, he keeps tossing more syntax layers on top of Swift.

Swift

Swift amazes me. A beta language that breaks your code every 6 months, a type system so totalitarian and inescapable it makes BDSM Haskell look like a vacation (and apparently nobody’s read Gödel’s paper), the founder abandoned it to go work on cars, Apple won’t ship production code in it, compiling burns your fucking CPU to the ground for 10s of minutes for code C can do in seconds, and after 3 years Xcode still can’t refactor it.

And stupid motherfuckers write their production apps in it. 😦

I know Objective-C is hard. It’s C plus Smalltalk, both of which are subtle and take a year or two to learn. [brackets scare:theNoobs] && dot.syntax.isOverloaded;
But the tools fucking work. Dynamic code makes programmers efficient. A more elegant weapon for a more civilized age.