Menu
Search

Forums

Forum Index

Thread overview
Operator for pass by move
Dec 19, 2022
Per Nordlöw
Dec 19, 2022
Paul Backus
Dec 21, 2022
deadalnix
Dec 21, 2022
Timon Gehr
December 19, 2022
Gravatar of Per NordlöwPosted by Per NordlöwReply
Per Nordlöw
Gravatar of Per Nordlöw


Reply

Have anybody considered dedicating a specific operator for pass-by-move semantics in assignments and argument passing?

For instance,

y = x;
z = f(y);

copies x and y whereas

y = <<x;
z = f(<<y);

moves x and y.

December 19, 2022
Gravatar of Paul BackusPosted by Paul Backus
in reply to Per Nordlöw		Reply
Paul Backus
Gravatar of Paul Backus
Posted in reply to Per Nordlöw


Reply

On Monday, 19 December 2022 at 11:12:54 UTC, Per Nordlöw wrote:

>

Have anybody considered dedicating a specific operator for pass-by-move semantics in assignments and argument passing?

Is core.lifetime.move not good enough?

December 21, 2022
Gravatar of deadalnixPosted by deadalnix
in reply to Paul Backus		Reply
deadalnix
Gravatar of deadalnix
Posted in reply to Paul Backus


Reply

On Monday, 19 December 2022 at 17:43:12 UTC, Paul Backus wrote:

>

On Monday, 19 December 2022 at 11:12:54 UTC, Per Nordlöw wrote:

>

Have anybody considered dedicating a specific operator for pass-by-move semantics in assignments and argument passing?

Is core.lifetime.move not good enough?

There is no reason for either.

The compiler should definitively be able to move things when appropriate (if they are not reused after) and this is allowed by the current spec.

December 21, 2022
Gravatar of Timon GehrPosted by Timon Gehr
in reply to deadalnix		Reply
Timon Gehr
Gravatar of Timon Gehr
Posted in reply to deadalnix


Reply
On 12/21/22 17:50, deadalnix wrote:
> On Monday, 19 December 2022 at 17:43:12 UTC, Paul Backus wrote:
>> On Monday, 19 December 2022 at 11:12:54 UTC, Per Nordlöw wrote:
>>> Have anybody considered dedicating a specific operator for pass-by-move semantics in assignments and argument passing?
>>
>> Is `core.lifetime.move` not good enough?
> 
> There is no reason for either.
> 
> The compiler should definitively be able to move things when appropriate (if they are not reused after) and this is allowed by the current spec.

The reason is that you may want to be explicit about your moves and have the type checker complain if it's actually used again. (This is also why core.lifetime.move is not great.)