Edzer Pebesma
Your confusion is caused by recomputing of the bounding box: ```r library(sf) # Linking to GEOS 3.8.0, GDAL 3.0.4, PROJ 7.0.0 sf::st_bbox(c(xmin = -20, xmax = 30, ymin = 30,...
Closed through https://github.com/r-spatial/sf/commit/84698a6894915854119d2cd266544f850307b583
There's a Dockerfile [here](https://github.com/r-spatial/sf/blob/main/inst/docker/gdal/Dockerfile) that tries to do sth similar (but using cmake, which is what modern GDAL and PROJ versions are switching to); it might be that you have...
`st_convex_hull()` doesn't support spherical geometry, it won't take care of the dateline wrapping by itself.
For the area involved: add 180 to latitude, do convex hull, subtract 180. `st_shift_longitude()`?
@fdetsch a reverse dependency check would reveal whether any CRAN packages are using these functions.
Oh, I do this much more primitive, e.g. for `sf` I use ```sh rm -fr /tmp/rdeps mkdir /tmp/rdeps cp sf_*gz /tmp/rdeps (cd /tmp/rdeps; _R_CHECK_FORCE_SUGGESTS_=FALSE R --save -e 'r
Deprecate them, and write the deprecation date/version in a comment in the function when you do so. When you later run into it and the package has seen a few...
If a a 0 buffer returns a valid polygon, what would be the polygon nodes of you compute a zero buffer around a `POINT`? S2 makes a micro-polygon, so seems...
It's not failure, it's the only correct thing.