C++ is different from languages you’re used to

Unless the language you’re used to is C/Rust/etc.

This article does not teach C++ in all of its glory. Rather, it teaches one specific insight that will help you as you continue to learn all of the rest.

One short text, one footnote, one practice problem.

The short text:

You’re not writing C++ as a joke, you’re writing it because you have a problem with very specific requirements that are hard or impossible to solve in the other language you typically write, and so the heavy guns of C++ are called in.

And you run into some annoyance where C++ is weird, and you think “why not fix this? The obvious default is to just” - and: No. C++ will expect you to have specific opinions on how to solve problems, to not be satisfied with a 99% fine default solution, because if a default solution was fine, you’d use the other language.

That’s it, that’s the insight. C++ will expect you to control many things you don’t actually need to use for solving your problem today, because somebody will need them to solve some problem some day.

Some things can have good standardized solutions, but they cannot be turned into built-in solutions that everybody have to use. They have to go in the standard library, as one solution among many.

The footnote:

Not all weird things are weird for good reasons, some parts of C++ are just bad - it’s an old language, built on top of an even older language. But they didn’t know better and neither did you.

The practice problem:

Try to fill in the empty function in this program using only language built-ins, without any library features.

The test cases stop at string length 17, but the solution should be about as robust against long strings as the C# example solution.

Interactive editor, thank you Matt Godbolt: https://godbolt.org/z/q8TnYv44Y

#include <cassert>
#include <string>

std::string second_ordered_by_first(std::string);

int main() {
    assert(second_ordered_by_first("") == "");
    assert(second_ordered_by_first("a1 n1 a2 n2 a3 n3") == "n1 n2 n3");
    assert(second_ordered_by_first("a1 n1 a3 n3 a2 n2") == "n1 n2 n3");
    assert(second_ordered_by_first("a3 n3 a2 n2 a1 n1") == "n1 n2 n3");
    assert(second_ordered_by_first("a3 n2 a1 n3 a2 n1") == "n3 n1 n2");
}

// Your implementation starts here.
// You may find it helpful to create the following supporting features as you fill in the second_ordered_by_first
// - container structures
// - a sort function
// - a string class

std::string second_ordered_by_first(std::string str) {

    return str;
}

Note: This is not intended to be an easy practice problem. I expect you to stop before you reach the end, unless you spend a lot more time than makes sense. Instead, it is intended to shock you out of programming assumptions you may be bringing along from other languages, defaults you expect to just have, which C++ does not have at all. C++ doesn’t have a built-in string class, that’s how dire the situation is!

For reference, here’s an easy way to do it in C# if you’re not being careful about performance: https://godbolt.org/z/d86GM4zKn

static string SecondOrderedByFirst(string input)
{
        var values = input
            .Split(' ')
            .Chunk(2)
            .OrderBy(c => c.First())
            .Select(c => c.Last());
        
        var first = values.First();
        
        return values
            .Skip(1)
            .Aggregate(first, (x, y) => x + ' ' + y);
}