geom_vline() does not extend the limits if the axis is a Date

[DanChaltiel]

Hi,

When using a date axis, geom_vline() will only show if there is data nearby the intercept.
For numeric axes though (y in my example below), if the intercept is off-limit, the plot will extend to show the line, which is the expected behavior.

library(tidyverse)
library(lubridate)

set.seed(1234)
df = tibble(
  x = as.Date(0:19, origin=ymd("2022-01-01")), 
  y=rnorm(20)
)
df %>% 
  ggplot(aes(x,y)) + 
  geom_point() + 
  geom_hline(yintercept=5) +                 #will show (extends)
  geom_vline(xintercept=ymd("2022-01-10")) + #will show (within limits)
  geom_vline(xintercept=ymd("2022-03-01")) + #will not show (off-limit without extension)

^{Created on 2022-08-25 with reprex v2.0.2}

Session info

sessionInfo()
#> R version 4.2.1 (2022-06-23 ucrt)
#> Platform: x86_64-w64-mingw32/x64 (64-bit)
#> Running under: Windows 10 x64 (build 14393)
#> 
#> Matrix products: default
#> 
#> locale:
#> [1] LC_COLLATE=French_France.1252  LC_CTYPE=French_France.1252   
#> [3] LC_MONETARY=French_France.1252 LC_NUMERIC=C                  
#> [5] LC_TIME=French_France.1252    
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> other attached packages:
#>  [1] lubridate_1.8.0 forcats_0.5.2   stringr_1.4.1   dplyr_1.0.9    
#>  [5] purrr_0.3.4     readr_2.1.2     tidyr_1.2.0     tibble_3.1.8   
#>  [9] ggplot2_3.3.6   tidyverse_1.3.2
#> 
#> loaded via a namespace (and not attached):
#>  [1] assertthat_0.2.1    digest_0.6.29       utf8_1.2.2         
#>  [4] mime_0.12           R6_2.5.1            cellranger_1.1.0   
#>  [7] backports_1.4.1     reprex_2.0.2        evaluate_0.16      
#> [10] httr_1.4.4          highr_0.9           pillar_1.8.1       
#> [13] rlang_1.0.4         curl_4.3.2          googlesheets4_1.0.1
#> [16] readxl_1.4.1        rstudioapi_0.14     R.utils_2.12.0     
#> [19] R.oo_1.25.0         rmarkdown_2.16      styler_1.7.0       
#> [22] labeling_0.4.2      googledrive_2.0.0   munsell_0.5.0      
#> [25] broom_1.0.0         compiler_4.2.1      modelr_0.1.9       
#> [28] xfun_0.32           pkgconfig_2.0.3     htmltools_0.5.3    
#> [31] tidyselect_1.1.2    fansi_1.0.3         crayon_1.5.1       
#> [34] tzdb_0.3.0          dbplyr_2.2.1        withr_2.5.0        
#> [37] R.methodsS3_1.8.2   grid_4.2.1          jsonlite_1.8.0     
#> [40] gtable_0.3.0        lifecycle_1.0.1     DBI_1.1.3          
#> [43] magrittr_2.0.3      scales_1.2.1        cli_3.3.0          
#> [46] stringi_1.7.8       farver_2.1.1        fs_1.5.2           
#> [49] xml2_1.3.3          ellipsis_0.3.2      generics_0.1.3     
#> [52] vctrs_0.4.1         tools_4.2.1         R.cache_0.16.0     
#> [55] glue_1.6.2          hms_1.1.2           fastmap_1.1.0      
#> [58] yaml_2.3.5          colorspace_2.0-3    gargle_1.2.0       
#> [61] rvest_1.0.3         knitr_1.40          haven_2.5.1

A workaround would be to add geom_blank(aes(x=ymd("2022-03-01"))) but it feels a bit unnecessary.

[teunbrand]

I imagine that is due to scale_x_date()$aesthetics not including xintercept. In attempt to avoid Chesterton's fence: is there a reason that some position scales don't default to ggplot2:::ggplot_global${x/y}_aes?

[thomasp85]

I believe this was attempted leading up to the latest release but rolled back because it caused a slew of unexpected issues... but can't remember the PR on top of my head