@jeroen this is a sequencing problem. https://r-spatial.github.io/evolution/ describes the background - we're now down to ~ 160 CRAN packages failing in one way or another when using sp 2.1-0 (to be released late September/early October) which uses sf if available, and uses none of the retiring packages rgdal, rgeos or maptools and when the retiring packages are not on the library path. About 120 of these failing packages have github issues, the rest were alerted up to 8 months ago by email.

Should I tell you when the retiring packages have been archived on CRAN, ao that your processes can stop tracking R-forge? The R-forge repos will remain in place, as a resource of last resort to any users who get stuck, and to provide minimal protection in case security bugs are found in PROJ, GDAL or GEOS, but no support for users.

Small extra point, rgrass7 will be fully replaced by rgrass and archived also in early October, as rgrass7 uses rgdal but rgrass uses terra.