how about base_view performance

[syz3843486]

template<std::size_t Curr, typename Func, std::size_t... Index>
    void each(Func func, std::index_sequence<Index...>) const {
        for(const auto curr: std::get<Curr>(pools)->each()) {
            const auto entt = std::get<0>(curr);

            if(((sizeof...(Get) != 1u) || (entt != tombstone))
               && ((Curr == Index || std::get<Index>(pools)->contains(entt)) && ...)
               && std::apply([entt](const auto *...epool) { return (!epool->contains(entt) && ...); }, filter)) {
                if constexpr(is_applicable_v<Func, decltype(std::tuple_cat(std::tuple<entity_type>{}, std::declval<basic_view>().get({})))>) {
                    std::apply(func, std::tuple_cat(std::make_tuple(entt), dispatch_get<Curr, Index>(curr)...));
                } else {
                    std::apply(func, std::tuple_cat(dispatch_get<Curr, Index>(curr)...));
                }
            }
        }
    }

as i see,when each entities,travel all entities and then filter.If I understood correctly?
as i see,normally ecs has archetype to accelerate,but i do not find in entt.
thank you

[syz3843486]

maybe i should use group?

[skypjack]

I strongly suggest you to read my blog (skypjack.github.io) and the ecs back and forth series in particular.
It contains many details about the different ecs models and should make it clear why there doesn't exist a normally in this field. 🙂

[syz3843486]

thank you

[syz3843486]

thank you for your blog very much！
I gently suggest adding some pictures,it would helpful for understand the details.
anyway thank you!