animated ggplotly bar chart does not work anymore

[stla]

Hello,

library(plotly)
p <- mtcars %>% ggplot(aes(x = mpg, y = cyl, frame = hp)) + geom_bar(stat = "identity")
ggplotly(p)

This throws an error: Error in -data$group: invalid argument to unary operator. I'm using the latest versions of the packages, and R 3-5-3.

Previously that worked: I have tried with R 3-3-2, ggplot2 2-2-1 and plotly 4-6-0, and this works.

[cpsievert]

Huh, this is weird. For a temporary workaround, you can use the new position_dodge2() or position_identity():

library(plotly)
p <- ggplot(mtcars, aes(x = mpg, y = cyl, frame = hp)) + geom_col(position = "dodge2")
ggplotly(p)

[roboton]

The workaround kind of works. Example below sets position to dodge2 but each y value in each frame has a different x position within the same discrete x value. Example below:

library(tidyverse)
library(plotly)

p <- map(1:100, ~ tibble(name = c("A", "B", "A", "B"), type = c("C", "D", "D", "C"), val = runif(4), time = .x))  %>%
    bind_rows() %>%
    ggplot(aes(name, val, fill = type, frame = time)) + geom_col(position = "dodge2", alpha = 0.1)

ggplotly(p)

Both stack and dodge throw the same Error in -data$group: invalid argument to unary operator error.

Setting position to identity works but the bars meant to be distinguished by type are drawn over each other.

The very manual workaround I've used is using the identity position in the ggplot, setting barmode to dodge using layout() and then needing to edit the width of each frame and trace in the plotly object so that the bars aren't too wide. Using style(width = xx) only modifies the width for the first frame but the width goes back to being too wide in subsequent frames.

Surely there's a better way? :)

gp <- ggplotly(p) %>% layout(barmode = "dodge")
sapply(seq_along(gp$x$data), function(i) gp$x$data[[i]]$width <<- 1/3 )