Creating a new ggplot2 theme

dromano · June 1, 2024, 3:50pm

I had a similar reaction to yours, @igor , so I followed @Gabor's suggestion and posted a question on github:

Superfluous guidance in "Creating a new theme" section of ggplot2 packages vignette?

opened 02:32PM - 01 Jun 24 UTC

documentation

[Motivated by Posit Forum post [Creating a new ggplot2 theme](https://forum.posi…t.co/t/creating-a-new-ggplot2-theme/187065).] The [Creating a new theme](https://ggplot2.tidyverse.org/articles/ggplot2-in-packages.html#creating-a-new-theme) section of the [vignette](https://github.com/tidyverse/ggplot2/blob/main/vignettes/ggplot2-in-packages.Rmd) on using ggplot2 in packages creates a new theme called `theme_custom()` and advises using a wrapper to apply the new theme: ``` default_theme <- function() { theme_custom() } mpg_drv_summary2 <- function() { mpg_drv_summary() + default_theme() } ``` However, since `theme_custom()` is already a function, the reason for the wrapper in not clear. The motivation given is that: > [...i]f not, the theme object is stored in the compiled bytecode of the built package, which may or may not align with the installed version of ggplot2! but my understanding is that this shouldn't happen unless, say, the package contained a call to `theme_set()`. Is the wrapper superfluous, or am I missing something?

The immediate response was a similar reaction by one of the maintainers:

Honestly, I'm a little surprised by this recommendation, or at least the wording of it. If you have a theme_custom() constructor, you wouldn't need a default_theme() wrapper.

I think the important bit to take away is that whenever you make a theme, always actually contruct it by running the contents through the theme() function. If you're making a package, don't save an already constructed theme to a variable or to disk, but have a function that builds the theme from stratch (or builds on a pre-existing theme).

So I expect the documentation you linked will be corrected soon, and in the meantime, you can find an alternative explanation of the do's and don't's in the reply to my post.