Futureproof way to modify default aes for ggplot2 geoms

tidyverse
1

Hi,

im implementing a package to create ggplot2 graphs conforming to the companies CI.

I have identified two ways to achieve this goal, but I wonder which one is most futureproof way. Either one has drawbacks.

Firstly, simply use ggplot2::update_geom_defaults()and set the defaults when loading the package like this:

geoms_color_updates <- c("line", "point")
geoms_fill_updates <- c("point", "bar", "col")

purrr::walk(geoms_color_updates, ~update_geom_defaults(.,
                                                       aes(color =  utils_colors_company$graph[["green_ci"]]))
)

purrr::walk(geoms_fill_updates, ~update_geom_defaults(.,
                                                      aes(fill =  utils_colors_company$graph[["green_ci"]]))
)

What rubs me the wrong way is that a global state is used to modify function behaviour which is less then ideal: The standard geoms behave differently depending on the state of the global environment. Not something I like to do.

Secondly, create custom geoms

GeomLineCompany <- ggproto("GeomLineCompany", GeomLine,
  default_aes = aes(
    color = utils_colors_company$graph[["green_company"]],
    linewidth = 1,
    linetype = 1,
    alpha = NA
  )
)


#' geom_line_company
#'
#' @inheritParams ggplot2::geom_line
#' @return
#' @export
geom_line_company <- function(mapping = NULL, data = NULL, stat = "identity",
                          position = "identity", na.rm = FALSE,
                          show.legend = NA, inherit.aes = TRUE, ...) {
  layer(
    geom = GeomLineCompany,
    data = data,
    mapping = mapping,
    stat = stat,
    position = position,
    show.legend = show.legend,
    inherit.aes = inherit.aes,
    params = list(na.rm = na.rm, ...)
  )
}

This has the benefit of clearly telegraphing the user that this is the company CI geom and that it behaves differently to the standard one. This would be me preffered solution where it not for the, pardon my french, horrid state of OOP in R. In Python I would simply create a class that inherits from the standard geom and modify the color. Yes, thats what happens with the ggproto class here, but the function that utilises the geom, geom_line_company, cannot simply be wrapped since I cant inject the different geom via .... Though ggproto is sort of OOP the method to use the class isnt tied to it nor does it inherit from it.

I worry that future changes in ggplot2 will cause this approach to break. So whats the best solution?

2

Have you considered taking the vanilla plot and running it through a filter that will parse the plot object, identify transgressions and issue dire threats. With every new drop, you are going to have to examine changes for messing with internal standards, right?

3

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.