rand(Normal(0, BigFloat) vs. rand(Truncated(Normal(0, BigFloat), -BigFloat, BigFloat)

[dtxl]

For example,

ND = Normal(0, BigFloat(0.9494405736148620092025834791894757387152032108223086656526968847923188266391808634))
rand(ND)

delivers a random BigFloat number, but e.g.,

TND = Truncated(Normal(0, BigFloat(0.9494405736148620092025834791894757387152032108223086656526968847923188266391808634)), -BigFloat(0.59803458093499065738957049860890123908234890864982480984902834098902348590890823), BigFloat(0.78742907123879873489723984758971829378973485789476878139475891738475897345897981227))
rand(TND)

gives the ERROR: MethodError: no method matching randnt(::TaskLocalRNG, ::BigFloat, ::BigFloat, ::BigFloat)
The function randnt exists, but no method is defined for this combination of argument types.

Closest candidates are:
randnt(::AbstractRNG, ::Float64, ::Float64, ::Float64)

thank you.

[dtxl]

In Distributions.jl/src/truncated/normal.jl the function randnt is

function randnt(rng::AbstractRNG, lb::Float64, ub::Float64, tp::Float64)
local r::Float64
...

Is there a reason for using Float64 instead of Real?
thank you.