Data gets dropped in add_trace and consequently points have the wrong color if NA is present

Have a look at the following code:

library(plotly)
x = c(NA, 2, 3, 4, 5)
color = c(0, 0, 0.5, 1, 1)
# If you try to plot it this way, you get mismatched colors. The NA seems to have been silently dropped and then the wrong indices aligned. The colors of points 2 and 3 are now blue, of 4 is purple and of 5 is green.
plotly::plot_ly()|>
    plotly::add_trace(
      type = "scatter",
      mode = "markers",
      x = x,
      y = 0,
      marker = list(
        color = color,
        cmin = 0,
        cmax = 1,
        colorscale = list(list(0, "blue"), list(0.5, "purple"), list(1, "green"))
      )
    ) |>
    plotly::layout(
      showlegend = FALSE
    ) |>
    plotly::hide_colorbar() 

It produces a plot with wrong colors:

I think this is unacceptable behavior, especially since this is not what happens if you do it in the what I consider the classical way:

data.frame(x=x, color=color) |> 
  plot_ly(x=~x, color=~color, y=0, colors = c("blue", "purple", "green")) |> 
  hide_colorbar()

Exécutez l’exemple R fourni via plotly::add_trace et comparez les couleurs des marqueurs avec l’exemple data.frame. Suivez la manière dont add_trace gère les valeurs NA dans x et aligne le vecteur de couleurs des marqueurs, puis ajoutez une couverture de régression montrant que les points non-NA conservent leurs couleurs correspondantes.