expressions as tick label broken

[RCinbo]

This issue was first posted as a reply to this issue. I was asked to make a new issue.
This code worked perfectly a while ago.

library(ggplot2)
library(latex2exp)
ggplot(mtcars, aes(x = wt, y = mpg)) +
    geom_point() +
    scale_x_continuous(breaks = 3, labels = TeX("$\\alpha$"))

yields

Error in vec_size():
! x must be a vector, not a <latexexpression/expression> object.

In a previous issue on a similar problem for annotations in a ggplot, specifying output = characterin the TeX function and parse = TRUE in the annotate function works. However, this cannot solve the problem at hand here:

ggplot(mtcars, aes(x = wt, y = mpg)) +
    geom_point() +
    scale_x_continuous(breaks = 3, labels = TeX("$\\alpha$", output = character), parse = TRUE)

yields:

Error in scale_x_continuous(breaks = 3, labels = TeX("$\alpha$", output = character), :
unused argument (parse = TRUE)

I found a workaround since labelling tick marks with a vector still seemed to work. I add a second label to a tick that I do not show on the graph by setting axis limits:

ggplot(mtcars, aes(x = wt, y = mpg)) +
    geom_point() +
    scale_x_continuous(breaks = c(0, 3),
                       labels = TeX(c("",  "$\\alpha$")),
                       limits = c(1.5, 5.5))

Another workaround, as suggested by @yutannihilation is using unclass or use parse(text =) instead of latex2exp.

ggplot(mtcars, aes(x = wt, y = mpg)) +
    geom_point() +
    scale_x_continuous(breaks = 3, labels = unclass(latex2exp::TeX("$\\alpha$")))

[yutannihilation]

Thanks! I removed the annotation() part for readability.

[teunbrand]

I'd like to point out this problem only shows up when there is a length-1 label passed to latex2exp::TeX(). The problem doesn't occur when the labels are longer.

a <- latex2exp::TeX("$\\alpha$")
b <- latex2exp::TeX(c("$\\alpha$", "$\\beta$"))

vctrs::obj_is_list(as.list(a))
#> [1] FALSE
vctrs::obj_is_list(as.list(b))
#> [1] TRUE

^{Created on 2025-09-24 with reprex v2.1.1}

I'm not entirely convinced that ggplot2 should take a defensive posture with regards to this special case.

[yutannihilation]

Oh, thanks, curious. I see. Then, until we figure out what is the supposed behavior, I close my pull request for now.

[teunbrand]

I don't think latex2exp is to blame for this, there is just something funky about custom-class expressions in general.

a <- expression(alpha)
b <- structure(a, class = "foo")

is.expression(a)
#> [1] TRUE
is.expression(b)
#> [1] TRUE

vctrs::obj_is_list(as.list(a))
#> [1] TRUE
vctrs::obj_is_list(as.list(b)) # why not?
#> [1] FALSE

^{Created on 2025-09-25 with reprex v2.1.1}

[ldecicco-USGS]

I've also had a function that has worked for many years now break with the new updates:

library(ggplot2)

fancyNumbers <- function(n){
  nNoNA <- n[!is.na(n)]
  x <- gsub(pattern = "1e", replacement = "10^", x = format(nNoNA, scientific = TRUE))
  non_one <- gsub(pattern = "e+", replacement = "", x = format(x, scientific = TRUE))
  non_one <- sapply(strsplit(non_one, "+"), function(x) x[1])
  non_one <- ifelse(non_one == "1", NA, non_one)
  x <- gsub(pattern = "e+", replacement = " x 10^", x = format(x, scientific = TRUE))
  
  exponents <- as.numeric(sapply(strsplit(x, "\\^"), function(j) j[2]))
  
  base <- ifelse(exponents == 0, "1",
                 ifelse(exponents == 1,
                        "10", "10^"))
  
  base[!is.na(non_one)] <- paste0(non_one[!is.na(non_one)], "%*%", base[!is.na(non_one)])
  exponents[base == "1" | base == "10"] <- ""
  textNums <- rep(NA, length(n))
  textNums[!is.na(n)] <- paste0(base, exponents)
  
  textReturn <- parse(text = textNums)
  return(textReturn)
}

data <- data.frame(x = 1:1000,
                   y = runif(1000, min = 0.0001, max = 100))

ggplot(data = data, aes(x = x, y = y)) +
  geom_point() +
  scale_y_log10(labels = fancyNumbers)

Error in `scale_y_log10()`:
! `breaks` and `labels` have different lengths.

If I go back to an older version, I'd get:

[yutannihilation]

@ldecicco-USGS
We already fixed it in the dev version.

devtools::load_all("~/repo/ggplot2")
#> ℹ Loading ggplot2
#> Overwriting method +(<ggplot2::ggplot>, <ANY>)
#> 
#> Overwriting method +(<ggplot2::theme>, <ANY>)
#> 
#> Overwriting method convert(<ggplot2::ggplot>, <list>)

fancyNumbers <- function(n){
  nNoNA <- n[!is.na(n)]
  x <- gsub(pattern = "1e", replacement = "10^", x = format(nNoNA, scientific = TRUE))
  non_one <- gsub(pattern = "e+", replacement = "", x = format(x, scientific = TRUE))
  non_one <- sapply(strsplit(non_one, "+"), function(x) x[1])
  non_one <- ifelse(non_one == "1", NA, non_one)
  x <- gsub(pattern = "e+", replacement = " x 10^", x = format(x, scientific = TRUE))
  
  exponents <- as.numeric(sapply(strsplit(x, "\\^"), function(j) j[2]))
  
  base <- ifelse(exponents == 0, "1",
                 ifelse(exponents == 1,
                        "10", "10^"))
  
  base[!is.na(non_one)] <- paste0(non_one[!is.na(non_one)], "%*%", base[!is.na(non_one)])
  exponents[base == "1" | base == "10"] <- ""
  textNums <- rep(NA, length(n))
  textNums[!is.na(n)] <- paste0(base, exponents)
  
  textReturn <- parse(text = textNums)
  return(textReturn)
}

data <- data.frame(x = 1:1000,
                   y = runif(1000, min = 0.0001, max = 100))

ggplot(data = data, aes(x = x, y = y)) +
  geom_point() +
  scale_y_log10(labels = fancyNumbers)

^{Created on 2025-09-26 with reprex v2.1.1}