geom_hline() syntax inconsistency - cannot specify yintercept outside of aes when using additional aesthetics

[RichardJActon]

It seems that you cannot use additional aesthetics with geom_hline if you leave yintercept outside of the aes call, unlike say y in geom_line.

df <- data.frame(
    "x"=rep(c(1,2,3),2),
    "y"=rep(c(1,2,3),2),
    "g"=rep(c(TRUE,FALSE),each=3)
)

---

This does not work contrary to what you would expect from things like: geom_line(y=1,aes(colour=x))

ggplot(df,aes(x=x,y=y)) + 
    geom_point() +
    geom_hline(yintercept = 1.5,aes(colour=g)) +
    facet_wrap(~g)

This does work but seems counter intuitive. Especially if you wanted, for example, 2 hlines one with a fixed yintercept and one that varied when it would seem more natural to leave the fixed intercept outside of the aesthetic and the variable one in.

ggplot(df,aes(x=x,y=y)) + 
  geom_point() +
  geom_hline(aes(yintercept = 1.5, colour=g)) +
  facet_wrap(~g)

[yutannihilation]

When yintercept argument is provided, new data and mapping for the layer are created , so the plot-wide data and mapping will be ignored. This trickiness seems intentional, but I too feel that this is a bit counter intuitive...

ggplot2/R/geom-hline.r

Lines 12 to 17 in 09ff5a4

	# Act like an annotation
	if (!missing(yintercept)) {
	data <- data.frame(yintercept = yintercept)
	mapping <- aes(yintercept = yintercept)
	show.legend <- FALSE
	}

geom_vline() and geom_abline() do the same thing.

ggplot2/R/geom-vline.r

Lines 12 to 17 in 09ff5a4

	# Act like an annotation
	if (!missing(xintercept)) {
	data <- data.frame(xintercept = xintercept)
	mapping <- aes(xintercept = xintercept)
	show.legend <- FALSE
	}

ggplot2/R/geom-abline.r

Lines 83 to 91 in 09ff5a4

	# Act like an annotation
	if (!missing(slope) || !missing(intercept)) {
	if (missing(slope)) slope <- 1
	if (missing(intercept)) intercept <- 0
	data <- data.frame(intercept = intercept, slope = slope)
	mapping <- aes(intercept = intercept, slope = slope)
	show.legend <- FALSE
	}

[hadley]

This is deliberate, and I don't think it can be changed (or at least the cost of analysing how hard it would be to change without breaking existing code would be greater than the perceived benefit)

[RichardJActon]

What about a warning if both y|x|intercept/slope and an aes are provided to these functions indicating the aes will be ignored unless the y|x|intercept/slope's assignment is also placed in the aes?

[hadley]

That would be reasonable. Want to do a PR?

[renanxcortes]

Hello, guys,

Hello, is this issue going to be merged in master? I think I'm facing the same problem, but I'm not really sure how to deal with it (I even tried to declare the data, but it seems it didn't work):

library(tidyverse)
data <- read_csv('https://raw.githubusercontent.com/renanxcortes/inequality-segregation-supplementary-files/master/data_ggplot2_issue.csv')


# WORKS JUST FINE
data %>%
  ggplot(aes(x = Index, y = p_value, col = Type)) +
  geom_point(size = 2.5) +
  facet_wrap(~factor(approachs,levels = c("systematic", "evenness", "permutation", "systematic_permutation", "even_permutation"))) +
  theme(axis.text.x = element_text(angle=90, hjust=1)) 



# Brakes accusing 'approachs' not found :(
data %>%
  ggplot(aes(x = Index, y = p_value, col = Type)) +
  geom_point(size = 2.5) +
  facet_wrap(~factor(approachs,levels = c("systematic", "evenness", "permutation", "systematic_permutation", "even_permutation"))) +
  theme(axis.text.x = element_text(angle=90, hjust=1)) +
  geom_hline(yintercept = 0.05, linetype = "dashed")

Kind regards,
Renan

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/