coord_cartesian() fails without reversed limits with scale_y_reverse()

[JWiley]

In 3.3.0 (including latest dev version), there is odd behavior when combining coord_cartesian() with scale_y_reverse().

This is fine:

ggplot(mtcars, aes(mpg, hp)) + 
  geom_point() + 
  coord_cartesian(xlim = c(10, 35), ylim = c(50, 350))

This also is fine:

ggplot(mtcars, aes(mpg, hp)) + 
  geom_point() + 
  scale_y_reverse() + 
  coord_cartesian(xlim = c(10, 35))

This breaks. I'm inclined to see it as a bug, because even if you think I should reverse the lower and upper limit of ylim given the use of scale_y_reverse(), why should that error make the axis ticks, etc. disappear altogether for the y axis?

ggplot(mtcars, aes(mpg, hp)) + 
  geom_point() + 
  scale_y_reverse() + 
  coord_cartesian(xlim = c(10, 35), ylim = c(50, 350))

Interestingly, this is fine again.

ggplot(mtcars, aes(mpg, hp)) + 
  geom_point() + 
  scale_y_reverse() + 
  coord_cartesian(xlim = c(10, 35), ylim = c(350, 50))

[yutannihilation]

Sorry, I didn't notice this issue... Let me close this in favor of #3952.