Re: [PATCH 2/2] rust: init: add re-initialization functions

[Benno Lossin]

On 29.04.24 14:24, Gary Guo wrote:
> On Thu, 25 Apr 2024 21:34:44 +0000
> Benno Lossin <benno.lossin@xxxxxxxxx> wrote:
> 
>> Sometimes it is necessary to split allocation and initialization into
>> two steps. One such situation is when reusing existing allocations
>> obtained via `Box::drop_contents`. See [1] for an example.
>> In order to support this use case add `re_[pin_]init` functions to the
>> pin-init API. These functions operate on already allocated smart
>> pointers that contain `MaybeUninit<T>`.
>>
>> Signed-off-by: Benno Lossin <benno.lossin@xxxxxxxxx>
>> Link: https://lore.kernel.org/rust-for-linux/f026532f-8594-4f18-9aa5-57ad3f5bc592@xxxxxxxxx/ [1]
> 
> 
> I don't find the re_init name very intuitive. From the name I would
> imagine these functions be taking a `Box<T>` and a `impl Init<T, E>`,
> dropping the content and produces a `Box<T>` again.

I see your point, but if you look at the link [1] from above, you will
see that there such a function wouldn't be helpful.

> Would it make more to rename the existing functions to have `new` in
> their name to indiciate that they allocate, e.g. `pin_new`, and have
> these functions that only does initialisation `init`/`pin_init`?

Since we now have full control over `Box::new` (via `BoxExt`), we could
also make it take a `impl Init<T, E>` instead of just `T`.
And we could also provide `fn pin(impl PinInit<T>) -> Pin<Box<T>>`.

I would happily rename the `re_init` functions to `init` in that case.
But if we don't want to do the other rename, then I think it would be
confusing to have the functions `new(T)`, `pin(T)`,
`pin_new(impl PinInit<T, E>)` and `new_in_place(impl Init<T, E>)`...

--
Cheers,
Benno