v0.5.0 does not add new fill scale in some cases

[owenssam1]

I had an issue with the newest version adding a fill scale, but when I reverted to 0.4.10, the issue went away. I was attempting to add a new fill scale via the code below, which had worked before I updated the package and works now that I have reverted.

Here is the error I was receiving:

"Error in $<-.data.frame(*tmp*, ".label", value = c("low", "moderate", : replacement has 4 rows, data has 3".

I was able to isolate the issue to the geom_point, which worked if I used fill outside the aes statement as just a single color. But as is, with the geom_point as the 2nd fill scale, the code will work up until the geom_point statement, but will not work if you try to include a second fill scale.

Here is the code I used:

 slf_binarized_2055_plot <- ggplot() +
      map_style +
      # data layer
      geom_raster(data = slf_binarized_2055_df, aes(x = x, y = y, fill = as.factor(global_regional_binarized))) +
      # add province layer
      geom_sf(data = locality_sf_plot_layer, aes(geometry = geometry), fill = NA, color = "black", linewidth = 0.3) +
      # fill scale raster
      scale_discrete_manual(
        name = "projected risk",
        values = values.obj,
        breaks = breaks.obj,
        labels = labels.obj,
        aesthetics = "fill",
        guide = guide_colorsteps(frame.colour = "black", ticks.colour = "black", barwidth = 20, draw.ulim = TRUE, draw.llim = TRUE)
      ) +
      # new scale
      ggnewscale::new_scale_fill() +
      # IVRs
## ISSUE STARTS HERE
      geom_point(data = IVR_locations_plot_layer, aes(x = x, y = y, fill = "viticultural\narea"), size = 2.5, shape = 21) +
      # fill scale for points
      scale_fill_manual(name = "", values = c("viticultural\narea" = "purple3")) +
      # aesthetics
      guides(fill = guide_legend(ncol = 1, byrow = TRUE, override.aes = list(size = 3))) +
      coord_sf()

[eliocamp]

Thanks for the report. In your example code your using datasets that I don't have access to, so I cannot reproduce the problem. Could you create a minimal reproducible example that uses as few extraneous objects as possible? You can either use build-in dataset or a minimal dataset that you can share here.

[owenssam1]

Sure! here are some minimal datasets:

library(terra)
library(rnaturalearthhires)

slf_binarized_1995_df <- data.frame(
  x = c(-156.6251, -156.5418, -156.4585, -156.3751, -156.8751),
  y = c(71.37486, 71.37486, 71.37486, 71.37486, 71.29153),
  global_regional_binarized = rep(5, 5)
)


IVR_locations_plot_layer <- data.frame(
  x = c(-123.61764, -123.53982, -123.46515),
  y = c(40.92875, 38.95189, 39.0750)
)

As for the locality_sf_plot_layer, you can use the package rnaturalearth to access the admin_1_states_provinces dataset: 

locality_sf <- rnaturalearth::ne_load(
      scale = 10,
      type = "admin_1_states_provinces",
      category = "cultural",
      destdir = ,
      returnclass = "sf"
    ) %>%
dplyr::filter(name == `United States`) # I do not actually remember the exact spelling of the USA in this dataset

Hope this helps

[eliocamp]

This might be a problem unrelated to ggnewscale. I get the same error just trying to plot the sf object.

library(ggplot2)

locality_sf <- rnaturalearth::ne_download(
  scale = 10,
  type = "admin_1_states_provinces",
  category = "cultural",
  returnclass = "sf"
) |> 
  dplyr::filter(name == "United States") 

ggplot() +
  geom_sf(data = locality_sf)
#> Error in `geom_sf()`:
#> ! Problem while setting up geom aesthetics.
#> ℹ Error occurred in the 1st layer.
#> Caused by error in `$<-.data.frame`:
#> ! replacement has 1 row, data has 0