Can't replicate plot in plotly 4.5.6 (from plotly 3.6.0)

[shamshmk]

I’m having trouble replicating a plot I created in R using plotly 3.6.0.
I recently installed plotly 4.5.6, and I’ve tried various ways to get the plot to work as before. I’m hoping you could give some insight on what I’m missing.

Code in version 3.6.0 :

x1 <- data.frame(x=rnorm(100,2),Name="x1")
x2 <- data.frame(x=rnorm(100,6),Name="x2")
x3 <- data.frame(x=rnorm(100,4),Name="x3")
x<-rbind(x1,x2,x3)
y <- rnorm(300)
df <- data.frame(x,y)

cols <- RColorBrewer::brewer.pal(length(unique(df$Name)), name = "Set1")
m1 <-list(size = 10,
          color = factor(df$Name, labels = cols[1:length(unique(df$Name))]),
          line = list(color = 'rgba(0, 0, 0, .8)', width = 1))

plot_ly(df,x = df$x,y=df$y, type="scatter",mode="markers", xaxis = "x1",
        yaxis = "y1",group = Name, marker = m1, legendgroup = Name) %>% 
add_trace(x = df$x, type="histogram",showlegend = F,xaxis = "x1",
              yaxis = "y2",group = Name, marker = m1, legendgroup = Name) %>% 
  add_trace(y = df$y,  type="histogram",showlegend = F,xaxis = "x2",
              yaxis = "y1",group = Name, marker = m1, legendgroup = Name) %>% 
  layout(
                                   showlegend = TRUE, 
                                   barmode="overlay",
                                   xaxis = list(anchor = 'y', domain = c(0, 0.8)),
                                   yaxis = list(anchor = 'x', domain = c(0, 0.8)),
                                   xaxis2 = list(anchor = 'y', domain = c(0.81, 1)),
                                   yaxis2 = list(anchor = 'x', domain = c(0.81, 1)))

this is the result:

I’m able to deselect a “trace” and it will affect the show/hide value for all three plots, for example if I click the blue “x2” this is the results:

I’ve tried doing this in plotly 4.5.6

plot_ly(df,x = ~x,y=~y, type="scatter", xaxis = "x1",
        yaxis = "y1",split = ~Name, marker = m1, legendgroup = ~Name) %>% 
  add_histogram(x = ~x, showlegend = F,xaxis = "x1",
            yaxis = "y2",split = ~Name, marker = m1, legendgroup = ~Name) %>% 
  add_histogram(x = ~y, showlegend = F,xaxis = "x2",
            yaxis = "y1",split = ~Name, marker = m1, legendgroup = ~Name) %>% 
  layout(
    showlegend = TRUE, 
    barmode="overlay",
    xaxis = list(anchor = 'y', domain = c(0, 0.8)),
    yaxis = list(anchor = 'x', domain = c(0, 0.8)),
    xaxis2 = list(anchor = 'y', domain = c(0.81, 1)),
    yaxis2 = list(anchor = 'x', domain = c(0.81, 1)))

but the result looked different and the legends would only control the scatter plot:

I was able to get it to look similar using subplots, but still the legend only controls the scatterplot:

s <- subplot(
  plot_ly(df,x = ~x, type="histogram",showlegend = F, marker = m1,color = ~factor(Name)),
  plotly_empty(), 
  plot_ly(df,x = ~x,y=~y, type="scatter",mode="markers",  marker = m1,color = ~factor(Name)), 
  plot_ly(df,y = ~y,  type="histogram",showlegend = F, marker = m1,color = ~factor(Name)),
  nrows = 2, heights = c(0.2, 0.8), widths = c(0.8, 0.2), 
  shareX = TRUE, shareY = TRUE, titleX = FALSE, titleY = FALSE
)
layout(s, showlegend = T, barmode = "overlay")

Any help would be greatly appreciated.

Thanks,
Shams

[cpsievert]

Thanks for writing in, and sorry for the long wait. 4092c1a provides a mechanism a way to address this issue (i.e., anchoring multiple traces to a single legend entry). See this test for my recommended way of approaching this use case