ggplot2 facet_wrap with two nested variables: use different strip.position/switch for each variable?

Question

I am making a ggplot where I use facet_wrap with two variables, say first var_1 with values A/B, second var_2 with 1/2, resulting into four facets, A-1, A-2, B-1, B-2. I would want the var_1 A/B title to show on top, and the var_2 1/2 to show up on the side. Doing this with facet_wrap(~dim_1 +dim_2, ncol=1), it seems however the argument strip.position will apply the same position for both? Another approach could be to use facet_grid() (which does want I want, to have facets both horizontal and vertical) but does not allow to set the number of rows/columns!?

Code below, with actual output, and what I want to have!

library(ggplot2)

df <- data.frame(dim_1 = rep(c("A", "B"), each=4),
                 dim_2 = rep(c("1", "2"), times=4),
                 x = rnorm(8), y = rnorm(8))

p <- ggplot(aes(x=x, y=y), data=df) +
  geom_point()


p+
  facet_wrap(~dim_1 +dim_2, ncol=1, strip.position = c("left"))

^{Created on 2019-09-29 by the reprex package (v0.3.0)}

Answer 1

I don't know a way to accomplish that within ggplot itself, but another approach would be to combine two plots using patchwork:

library(patchwork)
myplot <- function(section = "A") {
  ggplot(subset(df, df$dim_1 == section), aes(x,y)) + 
    geom_point() +
    # coord_cartesian(xlim = c(min(df$x), max(df$x))) +   # To get single x scale between all plots
    facet_grid(dim_2~dim_1, switch = "y") 
}
myplot("A") / myplot("B")