Multicategory bar plot failing with colour

Plotly R, MATLAB, Julia, Net
1

I am looking into the xaxis = list(type = multicategory) layout option for a bar plot; not having the most luck.

Consider the following reprex

df <- 
  tibble::tibble(
    cat1 = c("A", "B", "B", "C"),
    cat2 = c("N/A", "5", "10", "5"),
    value = 1:4
  )

df |> 
  plotly::plot_ly(
    x = ~ list(cat1, cat2),
    y = ~ value,
    color = ~ cat1, # Adds color to bars but breaks the plot
    type = "bar"
  ) |>
  plotly::layout(xaxis = list(type = "multicategory"))

When the color = ... line is omitted, the plot acts as expected with nested categories (A:B then N/A,5,10).
However, by adding color only two bars (both have cat1 = B) is shown and I have no idea why.

Any advice or indication where Iā€™m going wrong would be appreciated.

2

Used the dash app to investigate the data.

Seems like using list results in data like [ "A", "N/A"] for singley-observed groups and [ ["B", "B"], ["5", "10"]] otherwise.

My initial use of rbind meant the entire data was passed for every group and some value (such as "5") were put in the wrong cat1 (say, "A").

My solution is to add a trace for each category 1 / colour. That is,

plotly::plot_ly() |>
  plotly::add_trace(
    data = df[1, ],
    x = ~ rbind(cat1, cat2),
    y =  ~ value,
    color = ~ cat1, # Adds color to bars but breaks the plot
    type = "bar"
  ) |>
  plotly::add_trace(
    data = df[2:3, ],
    x = ~ rbind(cat1, cat2),
    y =  ~ value,
    color = ~ cat1, # Adds color to bars but breaks the plot
    type = "bar"
  ) |>
  plotly::add_trace(
    data = df[4, ],
    x = ~ rbind(cat1, cat2),
    y =  ~ value,
    color = ~ cat1, # Adds color to bars but breaks the plot
    type = "bar"
  ) |>
  plotly::layout(
    xaxis = list(type = "multicategory")
  )

I also installed library(dash) so may have updated plotly, pretty sure both reprex were run on Plotly 4.10.1 (R package version)