wrapping_$op() and saturating_$op() should be #[must_use]

[tormol]

I just had a bug due to a wrapping_add() having no effect because the return value was ignored:

let mut foo = foo();
foo.x.wrapping_add(bar); // should be foo.x = foo.x.wrapping_add(bar);
foo.y = baz; // so no warning about foo not needing to be mutable

It's an easy mistake to make when rewriting foo.x += bar to handle overflow.

EDIT: (I didn't do enough searching before opening this issue)

On one hand this is a subset of #48926, and covers a lot of methods so RFC #2450 would be better. On the other hand this issue is a v2 of #50124 for the parts of it not covered by #50149.

Meta

rustc 1.35.0-nightly (f8673e0ad 2019-04-03)
binary: rustc
commit-hash: f8673e0ad85e98997faa76fa7edc99c5825f46ee
commit-date: 2019-04-03
host: x86_64-unknown-linux-gnu
release: 1.35.0-nightly
LLVM version: 8.0

[tesuji]

What do you mean? If you comment the foo.y line, you would get the unused warning:
https://play.rust-lang.org/?version=nightly&mode=debug&edition=2018&gist=15090d92e645a5cdcd2af9071c92e3d9

[alex]

I would expect, in either version, a warning on the wrapping_add line about an unused result.

[varkor]

I agree that this could have been included as part of #50124, so I don't think it'd be controversial to open a pull request adding #[must_use] to the wrapping_* variants.

[memoryruins]

+1, I've seen this bite a few users in chats. searching logs, one said:

I'm using wrapping_add in yet another obligatory chip8 emulator and didn't store the result in a register, took me a few hours to find. thing is, I'm using that because of the panic for overflows, ironically exactly that is what caused me to write this bug

[Mark-Simulacrum]

AFAICT, this has since been done. Closing.