Suggestion from @EvaMaeRey: should we change make_constructor() to write_constructor()? I kind of like it because it writes a function for you so that you don't have to.

Hi Teun, Exciting to see the new function in action in the vignette!

I think the geom_chull case would not be one where I'd opt to write out the constructor instead of using make_constructor/write_constructor. I think the convention is arguably to expose the stat (even though it would be very strange to see a user switch this out).

Actually, in the ggplot2 book's example of geom_chull the stat argument is exposed, https://ggplot2-book.org/extensions.html#sec-modifying-geom-defaults, and I think this represents convention more than a locked-in stat (and will enable the super terse make_constructor move! ); maybe more representative of conventions?

And in the book case study, where geom_spring is first defined as a near alias to stat_spring, stat argument is also exposed https://ggplot2-book.org/ext-springs.html#constructors 21.2.4 Constructors And writing these two functions is a place where make_constructor/write_constructor really shines (otherwise scaffolding code 2x).

stat_spring <- make_constructor(StatSpring, geom = "path") # write a constructor instead of geom_path(stat = StatSpring)
geom_spring <- write_constructor(GeomPath, stat = "spring") # and just because 'ggplot2 users are accustomed to adding geoms, not stats, when building up a plot'

From base ggplot2, geom_count is another example where the stat is exposed, even though the function name kind of refers to the stat (StatSum)

ggplot2::geom_count

function (mapping = NULL, data = NULL, stat = "sum", position = "identity",
..., na.rm = FALSE, show.legend = NA, inherit.aes = TRUE)
{
layer(mapping = mapping, data = data, geom = "point", stat = stat,
position = position, show.legend = show.legend, inherit.aes = inherit.aes,
params = list2(na.rm = na.rm, ...))
}

While, I think there probably should be a couple of examples of write-it-out construction, I don't think geom_chull is the right spot to do it (so the vignette should probably follow the convention that geom_*() functions in general expose stat argument to users - as weird it would be for users to actually switch the stat out).