issue: position_dodge alters hover values

Problem: using position_dodge in at least geom_point seems to incorrectly assign the group values shown when hovering on a point. This is related to whether or not the values in the plot x-axis factor are alphabetically ordered.

Example:

library(ggplot2)
library(plotly)

#Data with x-axis factor in alphabetical order (a, b, c):
data <- data.frame('Col1' = c('a', 'b', 'c'), 'Col2'= factor(1:3), Col3 = c('A1', 'A2', 'A3'), 'Val' = c(10,5,1))
#Hovering on plot dots gives correct values, regardless of position_dodge.
ggplotly(ggplot(data, aes(x = Col1, y = Val, group = Col2:Col3)) + geom_point())
ggplotly(ggplot(data, aes(x = Col1, y = Val, group = Col2:Col3)) + geom_point(position = position_dodge(0.6)))

#Data with x-axis factor in non alphabetical order (c, a, b):
data1 <- data.frame('Col1' = c('c', 'a', 'b'), 'Col2'= factor(1:3), Col3 = c('A1', 'A2', 'A3'), 'Val' = c(10,5,1))
#Hovering on plot dots gives correct values if no position_dodge is used:
ggplotly(ggplot(data1, aes(x = Col1, y = Val, group = Col2:Col3)) + geom_point())
#Hovering on plot dots gives incorrect values (as if the x-axis levels had been reordered?) if position_dodge is used:
ggplotly(ggplot(data1, aes(x = Col1, y = Val, group = Col2:Col3)) + geom_point(position = position_dodge(0.6)))

Both solution and temporary workaround are appreciated. Plotly is superb, thank you so much for putting your time on it.

Comece executando os dois exemplos de ggplotly usando geom_point e position_dodge, comparando os casos alfabético e não alfabético de Col1. Rastreie como o ggplotly transfere dados de pontos agrupados para valores de hover; considera-se concluído quando os valores de grupo exibidos continuam correspondendo aos pontos corretos ao usar position_dodge.