Why I use base graphics instead of ggplot2

With base graphics, a histogram of 100 million points

d <- data.frame(x = rnorm(1e8))
system.time(hist(d$x))

takes

     user  system elapsed
    3.910   0.821   4.907

By contrast, if I use ggplot as in

library(ggplot2)
system.time({p<-ggplot(d) + aes(x=x) + geom_histogram();print(p)})

I see that the results take an order of magnitude longer:

  `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
     user  system elapsed
   35.308  13.481  50.338

As if an othe rder-of-magnitude speed reduction were not enough, there are some other things that I dislike.

1. The function ought to choose an appropriate binwidth, instead of saying that it has chosen a poor default.
2. The ggplot is in a low-contrast style can be a problem for some viewers and some printing/reproduction methods. (I know I don’t need to use the default, of course.)