ggplotly with "tooltip = text" is not working in a bar graph with + and - data

ggplotly with "tooltip = text" does not work in a bar graph, where there is positive and negative data.

It does work if a very small number is added x_limits[1] inside geom_col.

It also works if "tooltip = text" is removed

This does not work:

library(tidyverse)

plot_data <- ggplot2::diamonds %>%
  mutate(cut = stringr::str_to_sentence(cut)) %>%
  group_by(cut) %>%
  summarise(average_price = mean(price)) %>%
  mutate(average_price = round(average_price / 1000, 1)) %>% 
  mutate(average_price = average_price - 4) %>% 
  ungroup() %>% 
  add_row(cut = "OK", average_price = NA) %>% 
  mutate(tip_text = "Hello!")

na_data <- plot_data %>% 
  filter(is.na(average_price)) 

breaks <- pretty(plot_data$average_price)

limits <- c(min(breaks), max(breaks))

plot_data <- plot_data %>% 
  mutate(id = row_number())

plot <- ggplot(data = plot_data) +
  geom_col(aes(y = average_price, x = cut, text = tip_text))  +
  coord_flip() +
  scale_y_continuous(breaks = breaks, limits = limits, expand = c(0, 0)) +
  geom_col(aes(y = limits[2], x = cut, text = tip_text), data = na_data) +
  geom_col(aes(y = limits[1], x = cut, text = tip_text), data = na_data)

plot  

plotly::ggplotly(plot, tooltip = "text") 

This does works:

library(tidyverse)

plot_data <- ggplot2::diamonds %>%
  mutate(cut = stringr::str_to_sentence(cut)) %>%
  group_by(cut) %>%
  summarise(average_price = mean(price)) %>%
  mutate(average_price = round(average_price / 1000, 1)) %>% 
  mutate(average_price = average_price - 4) %>% 
  ungroup() %>% 
  add_row(cut = "OK", average_price = NA) %>% 
  mutate(tip_text = "Hello!")

na_data <- plot_data %>% 
  filter(is.na(average_price)) 

breaks <- pretty(plot_data$average_price)

limits <- c(min(breaks), max(breaks))

plot_data <- plot_data %>% 
  mutate(id = row_number())

plot <- ggplot(data = plot_data) +
  geom_col(aes(y = average_price, x = cut, text = tip_text))  +
  coord_flip() +
  scale_y_continuous(breaks = breaks, limits = limits, expand = c(0, 0)) +
  geom_col(aes(y = limits[2], x = cut, text = tip_text), data = na_data) +
  geom_col(aes(y = limits[1] + 0.000001, x = cut, text = tip_text), data = na_data)

plot  

plotly::ggplotly(plot, tooltip = "text") 

No source files or tests are named. Start by running the provided R example with plotly::ggplotly() and compare tooltip behavior with and without the small y-limit offset. Done means tooltip = "text" works for the bar graph containing both positive and negative data without requiring that workaround.