geom_hex no longer recognizes ..density.. in 2.1.0

[whao89]

In 2.1.0:

DF <- data.frame(x=rnorm(1000), y=rnorm(1000))
ggplot(DF) + geom_hex(aes(x=x, y=y, fill=..density..))

causes error:

Error in eval(expr, envir, enclos) : object 'density' not found

But in 2.0.0 (or other older versions), we get (random seeds notwithstanding):

[mikebirdgeneau]

I'm also running into this issue; it looks to be related to changes to the file R/stat-binhex.r in commit f7cb898

The previous code used to return:

# Convert to data frame
  data.frame(
    hexbin::hcell2xy(hb),
    count = hb@count,
    density = hb@count / sum(hb@count, na.rm = TRUE)
  )

whereas it now calls: hexBinSummarise() which ends up returning the results of hexbin::hcell2xy().

I'm not sure what the most elegant fix for this is to allow the previous functionality to remain.

[martinschmelzer]

I really like to have the old functionality back as well. Especially if you work with large samples and you want to plot gradient plots for all observations hexagon binning with alpha=..count.. was very useful! :)

[mikebirdgeneau]

This commit fixes it: https://github.com/mikebirdgeneau/ggplot2/commit/cf990038ffd66983a1ad2760f51c73049dcf56da

I'm not familiar with the reasons behind the change made by Hadley above, so I'm apprehensive to submit a pull request until I understand the implications.

I basically just added back these lines when preparing the data.frame output in hexBinSummarize():

out$count = as.vector(hb@count)
out$density = as.vector(hb@count / sum(hb@count, na.rm = TRUE))

[hadley]

@mikebirdgeneau looks like just an accidental omission. Can you please subset a PR?