Understanding plot recipes with identical arguments

General Usage
, ,
1

I just discovered the recipes mechanism in order to avoid having Plots as a package dependency.

I currently have a plot1(x::TypeX,y::TypeY) and plot2(x::TypeX,y::TypeY) functions which help present the same data in a different manner. Both have their advantages and shortcomings, so I would like to keep both available.

Now that I want to switch my definitions to recipes, I figured I could replace the

function plot1(x::TypeX,y::TypeY)
    <some plotting code>
end

definition to

@recipe function plot1(x::TypeX,y::TypeY)
    <some equivalent recipe code>
end

The same goes for plot2.

However, given the way recipes seem to works, the way to call this recipes is not plot1(X,Y) and plot2(X,Y), but simply plot(X,Y) for both, which makes one of the unavailable/overwritten.

I suppose I am not the first one to have this issue but I am confused… What is the correct way to avoid this. Using a @userplot maybe useful, but I do not see how it fits here.

Thanks !

2

Ok so I tested something with @userplot, which leads to

@userplot Plot1
@recipe function f(h::Plot1)
    if length(h.args) != 2 || !(typeof(h.args[1]) <: TypeX) || !(typeof(h.args[1]) <: TypeY)
        error("Wrong arguments, got $(typeof(h.args))")
    end
    <some recipe code>
end

Then I can call plot1(X,Y) as required. However, it feels weird to somewhat go around the whole multiple dispatch thing and validate the arguments afterwards. Can someone explain this approach ?