Layers inheriting aesthetics at the stage or whole aesthetic level

[mdlama]

Hi, everyone. I don't think there is an answer per se to this question, but I ran across an issue when I wanted to create the following side-by-side jitter and boxplot:

library(ggplot2)
ggplot(iris, 
       aes(x = Species, y = Sepal.Length)) + 
  geom_jitter(aes(x = stage(start = Species, after_scale = x - 0.1)), 
              width = 0.05) + 
  geom_boxplot(width = 0.1, 
               position = position_nudge(x = 0.1), 
               outlier.alpha = 0)

^{Created on 2021-05-24 by the reprex package (v0.3.0.9001)}

Basically, I wanted the plots to be centered on the tick marks, which required a nudge in opposite directions for each. The boxplot was easiest with the use of position_nudge. The stripchart was a little more difficult, given the fact that there is no built-in position_jitternudge. The fairly recent addition of our ability to manipulate the aesthetic mapping stages makes this possible, however, by specifying a manual nudge in the after_scale step. Notice, however, that I have to repeat the start aesthetic mapping, even though aesthetic inheriting is on for the geom_point layer (by default). To illustrate that I need to specify it, here's a reprex of just specifying the after_scale part (this was my initial try at the plot, where I incorrectly assumed the start stage would be inherited):

library(ggplot2)
ggplot(iris, 
       aes(x = Species, y = Sepal.Length)) + 
  geom_jitter(aes(x = after_scale(x - 0.1)), 
              width = 0.05) + 
  geom_boxplot(width = 0.1, 
               position = position_nudge(x = 0.1), 
               outlier.alpha = 0)
#> Error: geom_point requires the following missing aesthetics: x

^{Created on 2021-05-24 by the reprex package (v0.3.0.9001)}

In summary: it appears that aesthetic inheritance, and overriding them, happens at the whole mapping level, rather than at the individual stage level. So here's my question: Would it make sense to allow for overriding inherited aesthetics at the stage level to avoid duplicate mapping specifications?

[mara]

This is an interesting design question, and you might want to ask it as an issue in the ggplot2 repo, if there hasn't already been a related discussion.

Issues · tidyverse/ggplot2

[mdlama]

Thanks, Mara. I will indeed ask as an issue in the ggplot2 repo, and link to it here when it is up in case someone stumbles upon this in the future.

Edit:...and the issue on GitHub for the curious