ggplotly / barplot : Error in if (is.character(new.mode) && !is.na(new.mode))

[giacomovagni]

I am trying to plot a simple barplot with ggplot and ggplotly but I get this message error

Error in if (is.character(new.mode) && !is.na(new.mode)) { : missing value where TRUE/FALSE needed

I have no idea how to solve this issue.

Here is my ggplot code
a.
dta2 <- dta %>% melt %>% group_by(Var2, value) %>% summarise(count = n( )/nrow(dta))
b.
ggplot(dta2, aes(x = Var2, y = as.numeric(count), fill = factor(value))) + scale_fill_manual(values=rainbow(20)) + geom_bar(stat='identity', width = 1, position = "stack") + theme_minimal()
c.
ggplotly()

Error in if (is.character(new.mode) && !is.na(new.mode)) { : missing value where TRUE/FALSE needed

Any idea ?

Here is my reproductible data

dta = structure(c(7L, 2L, 6L, 48L, 56L, 7L, 7L, 7L, 46L, 7L, 7L, 24L, 7L, 19L, 7L, 30L, 7L, 24L, 42L, 67L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L, 7L, 2L, 27L, 48L, 56L, 7L, 7L, 7L, 67L, 7L, 7L, 24L, 7L, 20L, 7L, 30L, 7L, 67L, 42L, 23L), .Dim = c(20L, 10L), .Dimnames = list( c("1013662", "1024583", "1024812", "1025491", "1036642", "1037551", "1037552", "1037991", "1037992", "1040291", "1040292", "1041101", "1041212", "1041651", "104221", "104222", "104331", "104332", "104992", "1052571"), c("400", "401", "402", "403", "404", "405", "406", "407", "408", "409")))

[13bzhang]

Thanks for letting us know. I am looking into this.

[13bzhang]

I have been trying to debug this. The error happens when the count variable is not unique within a group. For instance:

# problem
testd <- data.frame(group1 = rep(1:3, each = 3),
                    group2 = LETTERS[1:3],
                    count = c(0.2, 0.2, 0.6)) # We have two 0.2 within each group.
p1 <- ggplot(testd, aes(x = factor(group1), y = count, 
                       fill = factor(group2))) + 
  geom_bar(stat='identity', width = 1, position = "stack")
ggplotly(p1)

This produces the error:

Error in if (is.character(new.mode) && !is.na(new.mode)) { : 
  missing value where TRUE/FALSE needed

In contrast,

# not a problem
testd <- data.frame(group1 = rep(1:3, each = 3),
                    group2 = LETTERS[1:3],
                    count = c(0.2, 0.1, 0.6))
p2 <- ggplot(testd, aes(x = factor(group1), y = count, 
                       fill = factor(group2))) + 
  geom_bar(stat='identity', width = 1, position = "stack")
ggplotly(p2)

produces the Plotly plot correctly:
https://plot.ly/~13bzhang/1444/count-vs-factorgroup1/

@cpsievert Could this be due to ggplotly's handling of stack?

[giacomovagni]

Thank you for your answer. I hope we can find a solution.

[cpsievert]

@13bzhang I have a vague idea of what's causing the problem. I think it has to do with some code I wrote a while back, so I'll plan on taking care of this (I might include it in #264 if it makes sense).

[hpneverback]

I have experience the same issue, thanks for finding the root cause @13bzhang , hope it will be fixed soon.