scale_*_steps (presumably binned_scales) doesn't like 0.01 with trans="log10"

[ldecicco-USGS]

I would like to bin my colors using a log10 transformation, and it seems that if the number 0.01 is included in the data, the scaling breaks with a warning:

Warning message:
In self$trans$transform(limits) : NaNs produced

This causes the transformation to happen, but the legend doesn't show up.

Here is an example of what doesn't work:

ggplot(data = data.frame(x = 1:5, 
                         y = c(0.01, 0.43, 0.46,
                               0.05, 0.17))) +
  geom_point(aes(x = x, y = y, color = y)) +
  scale_color_steps(trans = "log10")

Here are 3 examples that do work, which makes me suspect it is a bug:

ggplot(data = data.frame(x = 1:5, 
                         y = c(0.0101, 0.43, 0.46,
                               0.05, 0.17))) +
  geom_point(aes(x = x, y = y, color = y)) +
  scale_color_steps(trans = "log10") 

ggplot(data = data.frame(x = 1:5, 
                         y = c(0.00999, 0.43, 0.46,
                               0.05, 0.17))) +
  geom_point(aes(x = x, y = y, color = y)) +
  scale_color_steps(trans = "log10")

ggplot(data = data.frame(x = 1:6, 
                         y = c(0.01, 0.43, 0.46,
                               0.05, 0.17, 0.09))) +
  geom_point(aes(x = x, y = y, color = y)) +
  scale_color_steps()

I can speculate on why this is happening, but I'm unclear if there's an easy way to eliminate this problem.

One fix is to use "log1p" instead of "log10". It took awhile for me to pinpoint the issue and discover the solution. Even if a fix for 0.01 can't be included, it would be helpful to add some language in the binned_scale (or scale_*_steps) documentation.