`Error in vapply()` when trying to use `add_circle_layer()` on data containing lat/lon only

[etiennebacher]

Hello, I was trying to use add_circle_layer() to add a few points on a map and noticed that it would error if the data contained at least two rows and only columns for latitude and longitude.

Works (lat/lon + one column):

library(mapgl)

map_dat <- data.frame(lat = c(45, 50), lon = c(-45, 50), foo = 1:2) |>
  sf::st_as_sf(coords = c("lon", "lat"), crs = 4326) 

maplibre(projection = "mercator") |>
  add_circle_layer(id = "poi-layer", source = map_dat) 

Fails (lat/lon only):

library(mapgl)

map_dat <- data.frame(lat = c(45, 50), lon = c(-45, 50)) |>
  sf::st_as_sf(coords = c("lon", "lat"), crs = 4326) 

maplibre(projection = "mercator") |>
  add_circle_layer(id = "poi-layer", source = map_dat) 

# Error in `vapply()`:
# ! values must be length 1,
#  but FUN(X[[2]]) result is length 2

I don't use any other feature of the data in add_circle_layer() so this error looks like a bug to me.

---

mapgl 0.5.0

[Cidree]

@walkerke I was experiencing this same issue, so what I saw so far in case it can help:
Exploring the code, I see that geojsonsf::sf_geojson(source) returns the geojson by feature when source has no attributes, but the mapping needs a feature collection (i.e. geojsonsf::sf_geojson(source, simplify = FALSE) would resolve this issue).

When the source has attributes, the geojson is parsed correctly:

However, when there are no attributes, it should go with simplify = TRUE:

Therefore, I think forcing simplify = FALSE would solve this. However, I will raise this issue in geosjonsf as it seems the problem is here:

[Cidree]

Actually, reading the documenation of simplify it says the following:

logical indicating if sf objects without property columns should simplify (TRUE) into a vector of GeoJSON, or return a
Featurecollection with empty property fields (FALSE). If atomise is TRUE this argument is ignored.

Therefore, I think it has to be solved from the mapgl side.