adding scale_y_log10 changes y-axis to a weird range

[rxhu]

add_pvalue function affects scale_y_log10's behavior, e.g.

tg <- ToothGrowth
tg$dose <- as.factor(tg$dose)
tg$group <- factor(rep(c("grp1", "grp2"), 30))
two.means <- tibble::tribble(
~group1, ~group2, ~p, ~y.position,
"OJ", "VC", 0.0606, 36)

Without scale_y_log10, add_pvalue works as expected
ggplot(tg, aes(x = supp, y = len)) + geom_boxplot() + add_pvalue(two.means)

With scale_y_log10, y-axis become extremely large and original plots are squeezed to the bottom
ggplot(tg, aes(x = supp, y = len)) + geom_boxplot() + add_pvalue(two.means) + scale_y_log10()

[csdaw]

Thanks for the reproducible example! It seems to be a bug in the bracket y position calculation. Try installing from Github and it should fix the issue.

library(ggplot2)
library(ggprism)

tg <- ToothGrowth
tg$dose <- as.factor(tg$dose)
tg$group <- factor(rep(c("grp1", "grp2"), 30))
two.means <- tibble::tribble(
  ~group1, ~group2, ~p, ~y.position,
  "OJ", "VC", 0.0606, 36
)

ggplot(tg, aes(x = supp, y = len)) + geom_boxplot() + add_pvalue(two.means)

ggplot(tg, aes(x = supp, y = len)) + geom_boxplot() + scale_y_log10() + add_pvalue(two.means)

^{Created on 2022-06-27 by the reprex package (v2.0.1)}