stat_bin2d: doesn't handle integer-valued data

[tgs]

When I make a stat_bin2d using integer data, I still get the "range / 30" bin width.

> df <- data.frame(x = 0:3, y = 0:3)
> str(df)
'data.frame':   4 obs. of  2 variables:
 $ x: int  0 1 2 3
 $ y: int  0 1 2 3
> g <- ggplot(df, aes(x, y))
> g + stat_bin2d()

However, in stat-bin2d.r, it looks like it is trying to handle integer data specially, to end up with something like this:

> g + stat_bin2d(binwidth=c(1,1), origin=c(-0.5,-0.5))

Maybe, at some point, data$x retained its R integer type, but now something switches it to just numeric, and so the is.integer tests fail?

Here's a failing test case:

df <- data.frame(x = 0:3, y = 0:3)
g <- ggplot(df, aes(x, y))
p <- ggplot_build(g + stat_bin2d())

want_xbin <- cut(df$x, seq(-0.5, 4, 1))
expect_equal(p$data[[1]]$xbin, want_xbin)

[hadley]

Factors are turned into integers by the time the get to the stat function. It's not the best design, but it's too late to change now.