Boxplot legends

[cpsievert]

I'm not convinced we can/should provide legends for boxplots (see also #230). As far as I know, you can't draw multiple boxes in a single trace, so there must be a legend entry for unique x, making the legend redundant. For example, we could provide a legend for something like this:

library(plotly)
dat <- data.frame(
  cond = factor(rep(c("A", "B", "C", "D"), each = 200)), 
  col = factor(rep(c("C1", "C2"), each = 400)), 
  rating = c(rnorm(200), rnorm(200, mean = .8), rnorm(200, mean = .4), rnorm(200, mean = .2))
)
ggplot(dat, aes(x=cond, y=rating)) + 
  geom_boxplot(aes(fill=cond))

It'd be more useful to have a legend when color/fill is nested in x, but I don't think we could do something like this in plotlyjs, right @chriddyp?

ggplot(dat, aes(x=cond, y=rating)) + 
  geom_boxplot(aes(fill=col))

[cpsievert]

For now we'll go ahead and show a legend for each x value

[chriddyp]

hey @cpsievert - actually you can. Here's an example: https://plot.ly/~patsy.cline/35.embed. You have to include an x value for every single y value, i.e.:

{
    "x": [1, 1, 1],
    "y": [2, 1, 5]
}

Or, with multiple boxes (same trace so same color and same legend entry)

{
    "x": [1, 1, 1, 2, 2, 2],
    "y": [2, 1, 5, 3, 1, 5]
}