Implement `Default` for more types in the standard library

[tbu-]

Also add Hash to std::cmp::Ordering and most possible traits to
fmt::Error.

[rust-highfive]

r? @aturon

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

[tbu-]

There was a reddit comment that motivated this (especially for Mutex and RwLock), but I can't find it anymore.

[bluss]

The added other trait impls also should be mentioned in the commit message or PR.

[tbu-]

@bluss Updated.

[alexcrichton]

Thanks @tbu-! Tagging with T-libs to ensure that this comes up during triage.

[llogiq]

My question on StackOverflow has brought up the concern that a 0-duration is not a sensible default in most cases. Using this e.g. as a timeout is bound to fail at runtime. It's probably better not to implement Default for Duration at all.

[bluss]

Why would it fail at runtime? The drawbacks were not fully explained on SO.

[llogiq]

The problem is that Durations are often used as Timeouts and unless zero-duration timeouts are special case, the operation will always time out by default.

[tbu-]

I don't have strong feelings about this, we can definitely remove this if this isn't wanted. Two upsides: Using 0 seems to have some precedence (integers, floats), even though it's not clear where you'll use them. Implementing Default for more types means that you can #[derive] it in more cases.

[llogiq]

Yeah, that was exactly the argument I made in the reddit comment. And of course the default just has to make intuitive sense; we cannot foresee possible applications. Still, I'd like to have more input for this before we commit to an implementation, especially as insta-stability means we cannot easily back off.

[llogiq]

That reddit comment was by me. This would fix #31865 (and some more) by the way.

I'm also unsure if this should be directly stable.

[llogiq]

OK, so TIL trait impls are insta-stable. 😄

Apart from being unsure about the 'correct' default value for Duration (zero seems to make intuitive sense for timestamp arithmetic, but is a bad choice for timeouts), I'm in favor of this. It's more complete than #32807.

[alexcrichton]

The libs team discussed this during triage yesterday and the decision was that most of these seem fine but we should avoid doing something just for the sake of doing something. In line with that the default impls for Path and CStr seem like "just because", so can they be removed for now? Additionally, the Duration default seems like it's not "clearly the right choice" so can that be removed for now as well? (erring on the side of being conservative).

Other than that though this looked good to us!

[tbu-]

&CStr is pretty close to &str and Default is implemented for that type. Default is also implemented for &[T] and &mut [T].

The use case I have in mind here is when you have a Option<&CStr>, an unwrap_or_default should give you the empty string (like for Option<&str>).

[llogiq]

I also see no problem with CStr. Duration on the other hand should probably wait.

[alexcrichton]

Yeah splitting them out is fine, let's merge this and continue discussion on #32990

@bors: r+ 3df35a0

[bors]

⌛️ Testing commit 3df35a0 with merge 6fa61b8...

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-cargotest, auto-linux-64-debug-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-opt-mir, auto-linux-64-opt-rustbuild, auto-linux-64-x-android-t, auto-linux-64-x-armhf, auto-linux-64-x-armsf, auto-linux-64-x-freebsd, auto-linux-64-x-netbsd, auto-linux-cross-opt, auto-linux-musl-64-opt, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-mac-64-opt-rustbuild, auto-mac-ios-opt, auto-win-gnu-32-nopt-t, auto-win-gnu-32-opt, auto-win-gnu-32-opt-rustbuild, auto-win-gnu-64-nopt-t, auto-win-gnu-64-opt, auto-win-msvc-32-cross-opt, auto-win-msvc-32-opt, auto-win-msvc-64-cargotest, auto-win-msvc-64-opt, auto-win-msvc-64-opt-mir, auto-win-msvc-64-opt-rustbuild

[llogiq]

@tbu- 👍

[llogiq]

This fixes #31865.

[bombless]

I cannot think of a more proper default value for Duration other than zero though.

IMO if your scenario has a more proper value for a duration it should be a separate type and you may want to express duration itself through a trait.

And the implementation of Default might just be used casually, like https://github.com/bombless/to_default/blob/master/examples/simple.rs