Implement (part of) ACP 429: add DerefMut to Lazy[Cell/Lock]

[ChayimFriedman2]

DerefMut is instantly stable, as a trait impl. That means this needs an FCP.

@rustbot label +needs-fcp

rust-lang/libs-team#429

[rustbot]

r? @joboet

rustbot has assigned @joboet.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[ChayimFriedman2]

r​? libs-api

[joboet]

Weird, rustbot seems to be sleeping...
r? libs-api

[joboet]

The implementation looks good to me, r=me with the nit resolved after the FCP completes.

[SteveLauC]

Gentle ping on this PR, would love this to be merged soon:)

[ChayimFriedman2]

@SteveLauC This depends on a T-libs-api member starting a FCP.

[BurntSushi]

New APIs need to land as unstable. And an FCP isn't needed for that. There should also be a tracking issue. And the ACP should be linked.

[scottmcm]

Looks like this PR includes DerefMut, which is a trait impl, so that part does need an FCP.

Can you split this into two? One that's just T-libs and only adds unstable methods, and one that's T-libs-api with the stable trait impl. The former can land much more easily than the latter.

[ChayimFriedman2]

@BurntSushi @scottmcm This PR needs FCP for DerefMut and force_mut(), which, as I said in the OP, I decided to make instantly stable because it makes little sense to separate it from DerefMut.

For get_mut(), I already opened a tracking issue: #129333. I can split the PR though.

[workingjubilee]

DerefMut is instantly stable, as a trait impl; as such, I decided to also instantly stabilize force_mut(), since it makes little sense to have one but not the other. That means this needs an FCP.

so you're saying we could land force_mut() on nightly as unstable and then decide to stabilize DerefMut with it?

[ChayimFriedman2]

DerefMut is instantly stable, as a trait impl; as such, I decided to also instantly stabilize force_mut(), since it makes little sense to have one but not the other. That means this needs an FCP.

so you're saying we could land force_mut() on nightly as unstable and then decide to stabilize DerefMut with it?

That's an option, but that's not what I'm doing in this PR. Instead, I instantly stabilize both.

[bors]

☔ The latest upstream changes (presumably #130534) made this pull request unmergeable. Please resolve the merge conflicts.

[ChayimFriedman2]

@rustbot ready

I modified this to only add (immediately stable) DerefMut, after the other methods has been added unstably in #130476. This is now ready for FCP.

Although, libs-api may not want to include this until force_mut() stabilizes, or people may resort to this as a workaround (although I expect DerefMut to be used more often anyway, just like Deref vs. force()).