Date: Fri, 29 May 2020 09:38:50 +0200
I am not aware of the current status of reflection, but my hope would be
that hand-written type-erasure classes will be a thing of the past in
C++23/26. I don't have an opinion on whether we need a "unique_any" but
it would be a shame if we add it in C++23 and then in C++26 it could
have just been a typedef: "using unique_any =
type_class<no_member_functions, unique_storing>;"
Am 28.05.2020 um 15:41 schrieb Antony Polukhin via Std-Proposals:
> Hi,
>
> Is there a proposal for std::any like type that does not require held
> types to be copyable and has unique ownership of the held data?
>
> Is there interest in such class?
>
> We have an in house implementation that I find very useful for storing
> structures that contain unique_ptr (Flatbuffers, recursive types) and
> even non-movable types with mutexes.
>
that hand-written type-erasure classes will be a thing of the past in
C++23/26. I don't have an opinion on whether we need a "unique_any" but
it would be a shame if we add it in C++23 and then in C++26 it could
have just been a typedef: "using unique_any =
type_class<no_member_functions, unique_storing>;"
Am 28.05.2020 um 15:41 schrieb Antony Polukhin via Std-Proposals:
> Hi,
>
> Is there a proposal for std::any like type that does not require held
> types to be copyable and has unique ownership of the held data?
>
> Is there interest in such class?
>
> We have an in house implementation that I find very useful for storing
> structures that contain unique_ptr (Flatbuffers, recursive types) and
> even non-movable types with mutexes.
>
Received on 2020-05-29 02:41:57