Rfcs: `pushed` for returning a mutable reference to the pushed element.

[ticki]

It is annoying to manually fetch the last element, which is unsurprisingly quite common after a push. The most obvious thing would be to make Vec::<T>::push() return a &mut T, but this is strictly speaking a breaking change. It would be cool to get some crater data on whether this in practice is an issue (i.e. if people don't put the ; after push() in functions returning ()).

The more realistic change is to add a method, push_ref, returning &mut T.

[KalitaAlexey]

Hi @ticki,

Why not push_mut_ref according to conventions?

[comex]

I'd really like this too, but bikeshed: I don't like the name push_ref; it sounds like you're pushing a reference. Maybe pushed_ref?

[ticki]

@KalitaAlexey Two reasons:

1. The actual idiom is to use push_mut (like AsMut::as_mut).

2. Using _mut makes no sense when there is no non-mutuable alternative.

[ticki]

@comex what about pushed()? It's simpler and nicer IMO.

[comex]

@ticki Sure.

[nagisa]

The emplacement implementation could be made to return the reference to an element that has been just emplaced. I wonder, though, which is more useful – returning the index of the element or the reference to the element itself.

[ticki]

@nagisa

Returning a reference allows working around the implicit boundcheck, and honestly, getting the index isn't that hard (it's free too).

[eddyb]

IIRC, v.back() <- elem was supposed to produce a &mut T to the element.