Wolfgang Bangerth

So this is awkward, and I might need help from our vectorization specialists (@kronbichler @peterrum ?) I am trying something like this for the elements of a rank-1 tensor: ```...

Unlike `consteval`, there is no way whether the compiler is current evaluating a `constexpr` exprssion :-( For the constructor, though, one possible way to do that would be to have...

I misremembered the `consteval` thing. You *can* determine whether something is `constexpr` evaluated, but only since C++20: https://stackoverflow.com/questions/46919582/is-it-possible-to-test-if-a-constexpr-function-is-evaluated-at-compile-time

Hm, but this change already happened in #16474 and it compiled just fine...

Are we actually running into this with our current code base? #16474 did pass all of our tests, and as far as I know none of the periodic testers have...

I don't have any experience with GPUs or any of these compilers. Do we need to roll back all of these patches, or do you think you can finagle things...

I've given this some more thought. What I really want to do is this: ``` /** * Whether or not the 'values' array is stored as a vectorized array *...

Actually, I had thought about that and decided that it was too obscure an approach to suggest :-) But I had tried this and ran into a number of road...

`bit_cast` creates a *copy* of the object, whereas `reinterpret_cast` simply reinterprets a memory location :-( Either way, the problem with the non-`constexpr`ness of the intrinsics remains.

I played with this a bit more. In particular, I decided to remove a bunch of `constexpr` annotations in `Tensor` so that we can use `VectorizedArray` in the internal storage,...