Jonathan,

On Wed, Feb 23, 2022 at 2:48 PM Jonathan O'Connor via SG7 <
sg7_at_[hidden]> wrote:

> Matus,
> I've read your presentation and you explain why splicing needs to be done
> in a consteval context. However you don't show how to do splicing in your
> type-based approach. Having seen your recent Meetup talk, I remember you
> having get_value() and get_name() (?) functions which work outside of the
> consteval context. It might be useful to add a slide or two to clarify this
> point.
>

I was thinking about including a slide or two about that, but it's not
really relevant to the points I'm trying to make here.
As we've discussed at the Meeting C++ Q&A, I think that doing splicing by
both a generic expression (like [: ... :]) and by a set of named functions
like get_reflected_type, get_constant, get_reference, invoke, apply, etc.
is useful in various contexts, the former being more generic and the latter
more clear and readable. There is no big reason why we couldn't have both
regardless of how the metaobjects are represented and used as function
arguments.

--Matus