pumken/result

My implementation of a C++ Result<T, E> type inspired by Rust

Result

Overview

Result<T, E> is a template type that can be used to return and propage errors. It can be used to replace exceptions in context where they are not allowed or too slow to be used. Result<T, E> is an algebraic data type of Ok(T) that represents success and Err(E) representing an error.

Design of this class has been mainly inspired by Rust's std::result

struct Request {
};

struct Error {

    enum class Kind {
        Timeout,
        Invalid,
        TooLong
    }

    Error(Kind kind, std::string text);

    Kind kind;
    std::string text;
};

Result<Request, Error> parseRequest(const std::string& payload) {
    if (payload.size() > 512) return Err(Error(Kind::TooLong, "Request exceeded maximum allowed size (512 bytes)"));

    Request request;
    return Ok(request);
}

std::string payload = receivePayload();
auto request = parseRequest(payload).expect("Failed to parse request");

To return a successfull Result, use the Ok() function. To return an error one, use the Err() function.

Extract and unwrap

To extract the value from a Result<T, E> type, you can use the expect() function that will yield the value of an Ok(T) or terminate the program with an error message passed as a parameter.

Result<uint32_t, uint32_t> r1 = Ok(3u);

auto val = r1.expect("Failed to retrieve the value");
assert(val == 3);

unwrap() can also be used to extract the value of a Result, yielding the value of an Ok(T) value or terminating the program otherwise:

Result<uint32_t, uint32_t> r1 = Ok(3u);

auto val = r1.unwrap();
assert(val == 3);

Instead a terminating the program, unwrapOr can be used to return a default value for an Err(E) Result:

Result<uint32_t, uint32_t> r1 = Err(9u);

auto val = r1.unwrapOr(0);
assert(val == 0);

Map and bind

To transform (or map) a Result<T, E> to a Result<U, E>, Result provides a map member function. map will apply a function to a contained Ok(T) value and will return the result of the transformation, and will leave an Err(E) untouched:

std::string stringify(int val) { return std::to_string(val); }

Result<uint32_t, uint32_t> r1 = Ok(2u);
auto r2 = r1.map(stringify); // Maps a Result<uint32_t, uint32_t> to Result<std::string, uint32_t>

assert(r2.unwrap(), "2");

Note that map should return a simple value and not a Result<U, E>. A function returning nothing (void) applied to a Result<T, E> will yield a Result<void, E>.

To map a function to a contained Err(E) value, use the mapError function.

To bind a Result<T, E> to a Result<U, E>, you can use the andThen member function:

Result<uint32_t, uint32_t> square(uint32_t val) { return Ok(val * val); }

Result<uint32_t, uint32_t> r1 = Ok(3u);
auto r2 = r1.andThen(square);

assert(r2.unwrap(), 9);

Use orElse to apply a function to a contained Err(E) value:

Result<uint32_t, uint32_t> identity(uint32_t val) { return Ok(val); }

Result<uint32_t, uint32_t> r1 = Err(3u);
assert(r1.andThen(identity).orElse(square).unwrap(), 9);

The TRY macro

Like Rust, a TRY macro is also provided that comes in handy when writing code that calls a lot of functions returning a Result.

the TRY macro will simply call its argument and short-cirtcuit the function returning an Err(E) if the operation returned an error Result:

Result<void, IoError> copy(int srcFd, const char* dstFile) {

    auto fd = TRY(open(dstFile));
    auto data = TRY(read(srcFd));
    TRY(write(fd, data));

    return Ok();
}

Note that this macro uses a special extension called compound statement only supported by gcc and clang