Add NonEmptyVec::push

[jyn514]

And in general any methods that only allow adding to the vec but not removing. Otherwise it's really just a Box<NonEmptySlice> which doesn't seem to pull it's weight.

[tesuji]

Thanks for the interests.
I thought about adding NonEmptyVec::push or so but I don't particularly find the use case for it.

Also, push and insert are fallible operations, I am not sure how to cleanly handle them or just
mimic the std Vec API.

[jyn514]

When you say fallible you mean because the OS could run out of memory? It seems fine to panic in that case.

[tesuji]

And what about insert out of length ? I could return an Result<(), T> but it doesn't feel nice.

[tesuji]

How about adding a as_mut_vec -> &mut Vec? After all I don't know if the use case for push is big enough.

it is a bad idea. People could use pop on &mut Vec<T> and the whole contract by NonEmptyVec is gone.

[tesuji]

So I thought a bit and found that I need more evident about the use case for push.
If the OP has more than 10 thumps up, and there is a concrete example that push is semantically useful,
I could change my mind and add that API.

[jyn514]

I don't feel very strongly about it, it just seemed an obvious addition to the API.

[tesuji]

Yeah, the current inner implementation allows use easily adding push method to NonEmptyVec.
But just because we can does not make it through. It has to be useful unless we want APIs
that nobody uses.

Alright, back to your concern in the first post:

Otherwise it's really just a Box<NonEmptySlice> which doesn't seem to pull it's weight.

Maybe. But NonEmptyVec has inner capacity field, it should be really cheap to turn an instance of
NonEmptyVec<T> into Vec<T> and back, unlike Box<T>.