Cannot sample from a type that implements rand()

[nluetts]

Hi,

with regard to distributions from which to sample in a model, the Guide states:

distr can be a value of any type that implements rand(distr), which samples a value from the distribution distr.

https://github.com/TuringLang/Turing.jl/blob/d76d914231db0198b99e5ca5d69d80934ee016b3/docs/src/using-turing/guide.md?plain=1#L19

I have a set of samples from which I want to derive a distribution to sample from and use KernelDensityEstimate.jl to build a KDE from which one can sample using rand().
This returns a 1x1 Matrix, funny enough, but I wrapped the KDE in my own struct and defined the method Random.rand() such that it returns a float. It still does not work, I get the error:

ERROR: ArgumentError: the right-hand side of a `~` must be a `Distribution` or an array of `Distribution`s

From the Guide I was pretty sure my approach should work. Is the Guide wrong or is this a misunderstanding?
Thank you!

Turing v0.24.0, DynamicPPL v0.21.6, Julia 1.8.5