Unexpected behavior with NaN

[kylebutts]

I am not sure if this is intended behavior but NaNs now show up on the plot. Is this expected behavior? I would expect not if oob is censor?

library(ggplot2)
packageVersion("ggplot2")
#> [1] '4.0.0.9000'
ggplot() +
  geom_point(aes(x = c(1, 2, 3), y = c(1, 2, 1 / 0)))

^{Created on 2025-09-30 with reprex v2.1.1}

[EvaMaeRey]

Inf and -Inf are programmed to show up at the edge of the panel space, which can be helpful in some circumstances, but might also feel surprising.

See for example how -Inf is used here to draw to the edge of the panel space (minimum y). #6616 (comment)

An example of the maybe unwanted behavior...

ggplot() + 
  aes(x = 0:100, 
      y = 0:100) + 
  geom_point() + 
  scale_x_log10()

[teunbrand]

NaNs usually gets dropped for the same reasons NAs get dropped.
Like Gina says, what you created by 1 / 0 is an Inf, which is placed at the edge of the panel as they should.
If you want an even more censorious than oob_censor(), you can go for oob = scales::oob_censor_any() that will also throw out -Inf and Infs.
In any case, this is decidedly not a bug, so I'm going to close this issue.

[kylebutts]

Apologies. Bad reprex caused my own confusion!