Add non-panicking abs() functions to all signed integer types.

[jethrogb]

Rendered

[ExpHP]

Can you name a use case? It's hard for me to picture when it would be useful to receive a value of a different type when you want the absolute value of a number.

An alternative could be checked_abs(self) -> Option<Self> (which would use checked_neg for negatives).

---

Edit: I misread the proposal and thought it was suggesting that e.g. wrapping_abs(self: i32) -> u32. Now that I understand the proposal it feels rather like something that was simply "forgotten" from the standard library!

[jethrogb]

Obtaining the absolute value is going to always result in a positive number, so why would you not want to store that in an unsigned type? In any case, one example where this is useful for converting a number to a sign-and-magnitude representation:

(signed.is_negative(),signed.wrapping_abs() as u64)

The alternative you mention is not really an alternative but it is something that could be added to this RFC.

[jethrogb]

Added checked_abs and overflowing_abs to the RFC. I didn't add saturating_abs since there is no saturating_neg.

[glaebhoerl]

Rendered link seems to be broken

[glaebhoerl]

Earlier proposal for reference: Change abs() to return unsigned integers

[jethrogb]

Rendered link seems to be broken

Fixed

[brson]

checked_abs, not checking, would seem to be more consistent with existing conventions.

[jethrogb]

Yes, of course.

[brson]

Do all these have concrete use cases or are they just here to provide symmetry to other numeric methods? ISTM like checked_abs is the only one that really matters.

[jethrogb]

See above for a usecase for wrapping. I don't know of one for overflowing.

[alexcrichton]

🔔 This RFC is now entering its week-long final comment period 🔔

The libs team is leaning towards merging this as is, and we'd also be willing to just take a PR for this as well if you'd prefer to take that route @jethrogb :)

[jethrogb]

rust-lang/rust#35058

[alexcrichton]

Thanks @jethrogb! In that case I'll just close this in favor of that.