Use std::variant for polymorphic error

Accelerated Computing CUDA CUDA NVCC Compiler
1

I use std::variant achieve polymorphic in GPU. variant class up to 67 is ok, but 68 class in variant is error.
It shows type_traits(48): error class “std::_Conjunction<…>” has no member class "type.
In CPU, it’s ok to put so much class in variant.
CUDA version is 12.4 in Windows 10 x64, VS 2022.
Can I extend this variant?

2

cuda 12.6.2 still has this error.