primary attribute not taking effect in nested series

[juliohm]

Could you please confirm that the following should produce two series with the same color?

@userplot FooPlot

@recipe function f(foo::FooPlot)
  y = foo.args[1]
  x = 1:length(y)

  @series begin
    seriestype := :path
    primary := false

    x, y
  end

  seriestype --> :scatter
  
  x, 2y
end

fooplot(rand(3))

I remember using it in the past, but I am not sure.

[joshday]

I believe primary=false is used for removing a series from the legend, which is what is happening in your example. Questions like this may be better suited for slack or discourse.

[juliohm]

Thank you @joshday , I think this one is a real issue. I used to make use of the primary attribute for exactly that: maintain the previous style from the last plot command. It is not only about the legend. Check the example below, which still works as of today:

using Plots

x = linspace(0,1,10)

scatter(x)
plot!(x, primary=false)

The color is preserved with built-in plot commands. The issue is in defining nested series in plot recipes. If there is any other way to circumvent this, I'd be happy to learn.

[mkborregaard]

You're right @juliohm . I think this was possibly introduced by #1011 , but it's certainly an unwelcome side effect. Could you use this workaround?

@recipe function f(foo::FooPlot)
         y = foo.args[1]
         x = 1:length(y)

       @series begin
         seriestype --> :scatter
         x, 2y
       end


       @series begin
           seriestype := :path
           primary := false
           x, y
       end

       [],[]
end

[juliohm]

Thank you @mkborregaard, that is a very nice solution.

[juliohm]

@mkborregaard the solution works nicely, except that the legend is affected. Even though the second series copies the color and style, the legend shows a different series (for the [], [] pair?) in the final plot.

[mkborregaard]

Yes it seems to do that in some backends but not others. Just adding primary := false to the outermost series (`[],[]') should solve this.

[matthieugomez]

Still, it sounds like this should be solved. For instance, seriestype= :scatterpath now gives different colors for the dots and the lines, which was not the intent.

[mkborregaard]

maybe just a question of fixing the scatterpath recipe?

[matthieugomez]

Yeah, maybe. At least, it would help to show exactly what to do to go around this issue. In the meantime, could this issue be reopened?

…

On Fri, Sep 18, 2020 at 10:19 PM Michael Krabbe Borregaard < ***@***.***> wrote: maybe just a question of fixing the scatterpath recipe? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#1495 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABPPPXPPNLW477PXMM5YSFLSGQ5PZANCNFSM4E4CXXZQ> .

[matthieugomez]

Can be closed now #3423