consider geojsonsf

I see you're using geojsonio to convert sf objects to GeoJSON. Would you consider using geojonsf to improve speed?

Hopefully in this benchmark I'm replicating what you're doing inside topogram

library(microbenchmark)
library(geojsonsf)
library(geojsonio)

microbenchmark(
  
  geojsonsf = {
    shape$id <- 1:nrow(shape)
    sf_geojson( shape )
  },
  geojsonio = {
    geo_list <- geojson_list(input = shape)
    for (i in seq_along(geo_list$features)) {
      geo_list$features[[i]]$id <- i - 1
      geo_list$features[[i]]$properties$id <- i - 1
    }
    geo_json <- geojson_json(input = geo_list)
  },
  times = 5
  
)
# Unit: milliseconds
# expr        min         lq       mean     median         uq        max neval
# geojsonsf   1.955478   2.078431   2.172361   2.126749   2.226844   2.474302     5
# geojsonio 235.419525 239.354627 241.348199 240.610821 242.299392 249.056631     5

Note, I'm planning an update to geojsonsf by the end of October with some important fixes & updates.

---

You can also specify package version dependencies inside the DESCRIPTION in place of inside functions

  if (packageVersion("geojsonio") < "0.6.0.9100")
    stop("You need geojsonio >= 0.6.0.9100 to use this function.", call. = FALSE)

vs

Imports: 
    htmlwidgets,
    geojsonio (>=0.6.09100),
    htmltools,
    shiny,
    rmarkdown,
    jsonlite,
    magrittr

This way you don't have to implement the version check inside each function which may require it.

[pvictor]

That sounds very promising! I'll test it!

PS: the package version dependencies thingy is a lazy copy/paste from another package where it made more sense...