facet_wrap dir = "v" behavior changed

[EricGoldsmith]

In past versions of ggplot, the code below would render these facets vertically, not horizontally. The help docs imply that historic dir = "v" behavior was maintained - it has not been. I've tried various combinations of dir and as.table params to restore prior behavior, and have not been able to.

Here is the code to reproduce the bug:

library(tidyverse)

mtcars %>%
  ggplot(aes(x = mpg, y = carb)) +
    geom_line() +
    facet_wrap(vars(cyl), dir = "v")

Session info

R version 4.5.2 (2025-10-31)
Platform: aarch64-apple-darwin20
Running under: macOS Tahoe 26.4

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.1

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: America/New_York
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] lubridate_1.9.5 forcats_1.0.1   stringr_1.6.0   dplyr_1.2.1     purrr_1.2.1     readr_2.2.0     tidyr_1.3.2     tibble_3.3.1    ggplot2_4.0.2   tidyverse_2.0.0

loaded via a namespace (and not attached):
 [1] vctrs_0.7.2        cli_3.6.5          rlang_1.2.0        stringi_1.8.7      generics_0.1.4     S7_0.2.1           labeling_0.4.3     glue_1.8.0         hms_1.1.4          scales_1.4.0      
[11] grid_4.5.2         tzdb_0.5.0         lifecycle_1.0.5    compiler_4.5.2     RColorBrewer_1.1-3 timechange_0.4.0   pkgconfig_2.0.3    rstudioapi_0.18.0  farver_2.1.2       R6_2.6.1          
[21] tidyselect_1.2.1   pillar_1.11.1      magrittr_2.0.5     tools_4.5.2        withr_3.0.2        gtable_0.3.6       

[teunbrand]

Thanks for the report! Can you point me to the docs that say the historic dir = 'h' behaviour is preserved? I can't easily find this myself.

For those looking for a solution, the way to set 1-column layouts is to use ncol = 1.

[EricGoldsmith]

From ?facet_wrap, the description for var is:

Direction: either "h" for horizontal, the default, or "v", for vertical. When "h" or "v" will be combined with as.table to set final layout. Alternatively, a combination of "t" (top) or "b" (bottom) with "l" (left) or "r" (right) to set a layout directly. These two letters give the starting position and the first letter gives the growing direction. For example "rt" will place the first panel in the top-right and starts filling in panels right-to-left.

I read this to say historic behavior for "v" is maintained.

For those looking for a solution, the way to set 1-column layouts is to use ncol = 1.

This doesn't provide desired behavior for a dynamic number of facets (> 3), where facet_wrap determines the size of the resulting facet matrix (e.g. 3x4, etc.).

I've used dir = "v" for more than a decade, without issue. I would ask that its historic behavior be maintained.

[teunbrand]

I'm afraid we have a different readout of the docs then, as that is not the sense I'm distilling from the docs.
To revert this behaviour, the change needed will likely be in this bit:

ggplot2/R/facet-wrap.R

Lines 478 to 482 in c02c05a

	wrap_dims <- function(n, nrow = NULL, ncol = NULL) {
	if (is.null(ncol) && is.null(nrow)) {
	rc <- grDevices::n2mfrow(n)
	nrow <- rc[2]
	ncol <- rc[1]

Where we'd need to pass the dir argument and swap nrow/ncol when the growing direction is vertical (e.g. t* or b* direction).

[EricGoldsmith]

@hadley, would appreciate your thoughts here, on maintaining backward compatibility.

[hadley]

I have zero recollection of the dir argument, by my reading of the docs suggests it's supposed to control the direction of the layout within a 2d grid, not define the shape of the grid. So @EricGoldsmith I think you were relying on an off-label usage.

That said, the documentation for dir is pretty confusing, and I don't actually understand how the argument work because it seems to me that (e.g.) "tl" isn't a unique specification — yes the first cell goes at the top-left, but does it go left-to-right first or top-to-bottom? It's also confusing that it refers you to as.table which is superseded. So @teunbrand could you please propose some clearer docs?