Expand documentation of process::exit and exec

[nagisa]

Show a conventional way to use process::exit when destructors are considered important and also
mention that the same caveats wrt destructors apply to exec as well.

[rust-highfive]

r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

[steveklabnik]

@bors: delegate=kmcallister

r? @kmcallister

[bors]

✌️ @keeganmacallister can now approve this pull request

[bors]

✌️ @kmcallister can now approve this pull request

[frewsxcv]

What do you think about this as an example:

use std::io::{self, Write};
use std::process;

fn run_app() -> Result<(), ()> {
    // Your application logic here
    Ok(())
}

fn main() {
    process::exit(match run_app() {
        Ok(_) => 0,
        Err(_) => 1,
    });
}

[nagisa]

I personally prefer to not include an import for only one use of a function. I incorporated some of the other changes.

[frewsxcv]

Why are you printing the error if the error type is ()?

[frewsxcv]

I suppose I can see why, but it does seem unrelated to the point of this example. Not a big deal.

[frewsxcv]

Would it be more accurate to replace "any other thread's" with "child thread's"?

[nagisa]

No. “Child thread” makes sense in relation to “main thread”, but you can call process::exit on any thread you want.

[frewsxcv]

Is this platform specific behavior? I'm not familiar with how all the platforms handle execvp

[nagisa]

It is not. We always try our best to set CLOEXEC on all filedescriptors in the standard library and the behaviour of execvp wrt CLOEXEC fds is documented in posix: http://pubs.opengroup.org/onlinepubs/9699919799/functions/execvp.html

[frewsxcv]

Nit: that the execvp syscall

[frewsxcv]

Nit: compute the exit code

[nagisa]

Ping?

[frewsxcv]

I think the documentation uses American English, so this should be "behavior"

[alexcrichton]

@bors: r+

[bors]

📌 Commit c2eab73 has been approved by alexcrichton

[frewsxcv]

@bors rollup