Continuous fill does not obey fine-grained show.legend settings

[mjskay]

If you use fine-grained settings with show.legend, the legend for a continuous fill variable is not drawn. E.g. this:

library(ggplot2)
library(dplyr)

tribble(
  ~x1, ~y1, ~x2, ~y2, ~id, ~id_int,
    1,   1,   2,   2, "A",       1,
    2,   2,   3,   3, "B",       2
) %>%
  ggplot(aes(xmin = x1, ymin = y1, xmax = x2, ymax = y2, fill = id_int)) +
  # could also do show.legend = c(size = FALSE) here and get the same
  # result; in both cases I would have expected the fill legend to show up
  geom_rect(show.legend = c(size = FALSE, fill = TRUE))

^{Created on 2019-07-28 by the reprex package (v0.3.0)}

I chose size here as an unrelated variable, it could be anything. This problem also happens on geoms other than rect, so I don't think it's a problem there.

I would expect output that looks like this:

FWIW, with a discrete variable for the fill, it does work as expected:

library(ggplot2)
library(dplyr)

tribble(
  ~x1, ~y1, ~x2, ~y2, ~id, ~id_int,
    1,   1,   2,   2, "A",       1,
    2,   2,   3,   3, "B",       2
) %>%
  ggplot(aes(xmin = x1, ymin = y1, xmax = x2, ymax = y2, fill = id)) +
  # could also do show.legend = c(size = FALSE) here and get the correct result
  geom_rect(show.legend = c(size = FALSE, fill = TRUE))

^{Created on 2019-07-28 by the reprex package (v0.3.0)}

I tested on the github version of ggplot2 as of today (2019-07-28). Session info:

R version 3.6.0 (2019-04-26)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 18362)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

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

other attached packages:
 [1] reprex_0.3.0          gganimate_1.0.3       RColorBrewer_1.1-2    cowplot_1.0.0         forcats_0.4.0        
 [6] modelr_0.1.4          brms_2.9.0            Rcpp_1.0.2            broom_0.5.2           emmeans_1.3.5.1      
[11] tidybayes_1.1.0       rstan_2.19.2          StanHeaders_2.18.1-10 ggstance_0.3.2        ggplot2_3.2.0.9000   
[16] dplyr_0.8.3           magrittr_1.5         

loaded via a namespace (and not attached):
  [1] ggnewscale_0.3.0          TH.data_1.0-10            colorspace_1.4-1          ggridges_0.5.1           
  [5] rsconnect_0.8.15          rprojroot_1.3-2           estimability_1.3          markdown_1.0             
  [9] base64enc_0.1-3           fs_1.3.1                  rstudioapi_0.10           farver_1.1.0             
 [13] remotes_2.1.0             svUnit_0.7-12             DT_0.7                    mvtnorm_1.0-11           
 [17] dotwhisker_0.5.0          bridgesampling_0.7-2      codetools_0.2-16          splines_3.6.0            
 [21] knitr_1.23                shinythemes_1.1.2         zeallot_0.1.0             pkgload_1.0.2            
 [25] bayesplot_1.7.0           packrat_0.5.0             shiny_1.3.2               clipr_0.7.0              
 [29] compiler_3.6.0            backports_1.1.4           assertthat_0.2.1          Matrix_1.2-17            
 [33] lazyeval_0.2.2            cli_1.1.0                 tweenr_1.0.1              later_0.8.0              
 [37] htmltools_0.3.6           prettyunits_1.0.2         tools_3.6.0               igraph_1.2.4.1           
 [41] coda_0.19-3               gtable_0.3.0              glue_1.3.1                reshape2_1.4.3           
 [45] vctrs_0.2.0               nlme_3.1-140              crosstalk_1.0.0           xfun_0.8                 
 [49] stringr_1.4.0             ps_1.3.0                  mime_0.7                  miniUI_0.1.1.1           
 [53] gtools_3.8.1              devtools_2.1.0            MASS_7.3-51.4             zoo_1.8-6                
 [57] scales_1.0.0              colourpicker_1.0          hms_0.5.0                 promises_1.0.1           
 [61] Brobdingnag_1.2-6         parallel_3.6.0            sandwich_2.5-1            inline_0.3.15            
 [65] shinystan_2.5.0           curl_4.0                  memoise_1.1.0             gridExtra_2.3            
 [69] loo_2.1.0                 stringi_1.4.3             dygraphs_1.1.1.6          desc_1.2.0               
 [73] pkgbuild_1.0.3            rlang_0.4.0               pkgconfig_2.0.2           matrixStats_0.54.0       
 [77] evaluate_0.14             lattice_0.20-38           purrr_0.3.2               labeling_0.3             
 [81] rstantools_1.5.1          htmlwidgets_1.3           processx_3.4.1            tidyselect_0.2.5         
 [85] plyr_1.8.4                R6_2.4.0                  generics_0.0.2            multcomp_1.4-10          
 [89] whisker_0.3-2             pillar_1.4.2              withr_2.1.2               xts_0.11-2               
 [93] survival_2.44-1.1         abind_1.4-5               tibble_2.1.3              crayon_1.3.4             
 [97] arrayhelpers_1.0-20160527 rmarkdown_1.14            progress_1.2.2            usethis_1.5.1            
[101] grid_3.6.0                callr_3.3.1               threejs_0.3.1             digest_0.6.20            
[105] xtable_1.8-4              tidyr_0.8.3               httpuv_1.5.1              stats4_3.6.0             
[109] munsell_0.5.0             sessioninfo_1.1.1         shinyjs_1.0

[yutannihilation]

Thanks, confirmed. Probably we should check any(layer$show.legend[matched]) instead of isTRUE(layer$show.legend) around here:

ggplot2/R/guide-colorbar.r

Lines 242 to 248 in 141c775

	guide_geom.colorbar <- function(guide, layers, default_mapping) {
	# Layers that use this guide
	guide_layers <- lapply(layers, function(layer) {
	matched <- matched_aes(layer, guide, default_mapping)
	if (length(matched) > 0 && (isTRUE(is.na(layer$show.legend)) || isTRUE(layer$show.legend))) {
	layer

(I was afraid it was me that broke this by #3163, but this doesn't seem to work even before that...)

[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/