Simon Byrne
Simon Byrne
No, that was intentional! If you specify `mean` or `std` parameters, it will match the moments. ```julia julia> using Distributions julia> LogNormal(mu=1.0, sigma=1.0) LogNormal{Float64}(μ=1.0, σ=1.0) julia> LogNormal(mean=1.0, sigma=1.0) LogNormal{Float64}(μ=-0.5, σ=1.0)...
> I just wonder why `Normal(x, y)` should be allowed but not `Normal(x)`. If we assume people know the order of arguments (reason why we keep the two-argument method), omitting...
I would prefer to get rid of them to be honest: I was hoping that the "standard" constructor would be the keyword ones, and the positional ones would mainly be...
We can decide that later. If people feel particularly strongly, I can add some of the constructors back for now.
They're already part of the API. I'm just deprecating some I think are particularly confusing. (unless I'm misunderstanding what you're saying)
So I've made a few changes to KeywordDispatch.jl to make this a bit simpler. I've updated the normal distribution with the new format, and updated the help text: https://github.com/JuliaStats/Distributions.jl/pull/823/files#diff-490e3dc074dbfc901fe90e4163ed3048 I...
Hmm, it is slightly worrying that CI is timing out. I'll need to figure out what is going on.
Okay, that's better. Other than some tests are there any other changes people want to see?
Hmm, updating the tests is difficult due to the hardcoded R stuff.
@matbesancon are you referring to the CI timing out? It was because I was still calling a deprecated method in a loop. This has now been fixed.