Simplify reverse tuple iteration

[vittorioromeo]

Should help a bit with #1160. Would also be nice to see how it interacts with #1055

[fnc12]

@vittorioromeo thank you for this PR. Please resolve the conflicts to be able to run CI tests

[vittorioromeo]

@vittorioromeo thank you for this PR. Please resolve the conflicts to be able to run CI tests

Should be fine now :)

[fnc12]

@vittorioromeo did you test compilation time with your branch? I am actually not sure whether reverse tuple iteration is used anywhere so I am not sure whether your fix will make much effort

[vittorioromeo]

@vittorioromeo did you test compilation time with your branch? I am actually not sure whether reverse tuple iteration is used anywhere so I am not sure whether your fix will make much effort

I saw a very small improvement of average 24ms per instantiation of iterate_tuple from 28ms, but hard to tell if it's impactful in any way. There are a few calls to iterate_tuple<true> in the codebase. Regardless, I think that the PR is an improvement overall because of the removal of unnecessary code.

[trueqbit]

This is awesome C++! Thanks @vittorioromeo for thinking along.

I doubt that this change alone will improve compilation times overall.
But it simplifies the code, and that was what I was originally looking for when I updated the tuple iteration but couldn't find it!

By the way, there is good reading about this fold expression trickery:

• https://www.foonathan.net/2020/05/fold-tricks
• https://quuxplusone.github.io/blog/2020/05/07/assignment-operator-fold-expression/

[trueqbit]

BTW, we might implement first_index_sequence_value() as follows. What do you think?

template<size_t... Idx>
SQLITE_ORM_CONSTEVAL size_t first_index_sequence_value(std::index_sequence<Idx...>) {
    static_assert(sizeof...(Idx) > 0);
    size_t result;
    ((result = Idx, true) || ...);
    return result;
}

[vittorioromeo]

Compared to the current implementation, I'm not sure it's going to be faster. I would recommend a benchmark on https://www.build-bench.com/

[fnc12]

@vittorioromeo thanks!