On Mon, Nov 11, 2019 at 2:45 PM Corentin via SG7 <sg7@lists.isocpp.org> wrote:

Reading the minutes of the paper about consteval variables, it was not clear to me what the new direction was supposed to me but it seemed significant.

Could someone be kind enough to speak about what that new direction would offer that static reflection does not, what it would entail and what impact that may have on time frames?

To be clear, a new direction wasn't decided at the meeting. The item being revisited is how tightly coupled metaprogramming and the existing constexpr/consteval mechanisms will be. The constraints introduced by such a tight coupling make reasoning difficult in a lot of cases. The circle programming language (a C++ derivative) went with a different set of tradeoffs for metaprogramming and ended up with something arguably more compelling and simple from a user standpoint. It is also a lot faster at compile time.

We plan on doing a thorough investigation into this so we can clearly articulate what the tradeoffs are and what, if any, tweaks (hopefully) or fundamental changes (hopefully not) would be interesting to make with the current design before the next meeting.