Peter Dimov

I see how this has created a problem for you. Sorry. The way you have things set up, this probably appeared to work correctly, except that the optimized versions weren't...

> We have decided to port over a small subset of the old hashing implementation for our use: If you're going to do that, I suggest you use the new...

> You might also be interested in Aumasson and Bernstein's SipHash. It is meant to be a fast (AXR design), collision resistant with a uniform distribution. SipHash hashes a sequence...

Is `sizeof(decltype(D.name)) - 1` really a lot easier than `strlen(D.name)`? There's the problem with the latter not always being `constexpr`, sure, but if you need portability (to something that doesn't...

I show how a descriptor name is turned into a compile-time Hana string here: https://lists.boost.org/Archives/boost/2020/09/250046.php Not sure how your compile-time string works, but it should be similar.

I'm not 100% sure this won't create problems for a built-in compiler implementation down the road. I know that `__func__` is defined as an array and this causes trouble for...

I'll keep it open as a reminder. I've been looking into implementing Describe on top of the current reflection proposals ([P2320](http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2320r0.pdf) and [P1240](http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1240r1.pdf)) and at this point it's basically impossible...

`array` isn't good because it's also not possible to get a `string_view` or `char const*` that's enough-constexpr from it (because the `array` returned from a function has automatic storage and...

I'm familiar with PFR and that it recently got the ability to extract field names under C++20 (that was your contribution, right?) but even with the names, it still doesn't...

I mean, if you have ``` struct X { int x; }; BOOST_DESCRIBE_STRUCT(X, (), (x)) struct Y: X { int y; void f(); static int s; }; BOOST_DESCRIBE_STRUCT(Y, (X), (y,...