Edzer Pebesma

It's not that simple, but a PR to make this an option is welcome.

Welcome back! I see > In CPL_read_gdal(as.character(x), as.character(options), as.character(driver), : GDAL Message 1: Limiting number of bands to 32768 instead of 105144 wasn't there an environment variable that needed to...

Same issue here. What does work however is ```r > stars:::read_mdim("usafl_fg10_2007-2019.nc") stars object with 3 dimensions and 1 attribute attribute(s), summary of first 1e+05 cells: Min. 1st Qu. Median Mean...

See https://gdal.org/user/multidim_raster_data_model.html

Currently, `distinct.sf` works on the variables specified _and_ the geometry column; since that one, in your example, contains invalid geometries the error occurs when finding distinct geometries. A way out...

If you safe the object to `r`, then ```r > dim(st_dimensions(r)) grid_longitude grid_latitude time ensemble_member 418 406 3 1 > dim(r) grid_longitude grid_latitude time ensemble_member 100 100 3 1 ```...

btw raster will also ignore it if it is not noise, because it doesn't support rectilinear. But indeed: ```r > r= read_ncdf(f, ncsub = cbind(start = c(1, 1, 1, 1),...

I'd be OK. But also eps should have a better default than 1e-12.

This is what I can get: ```r f = "tas_rcp85_land-rcm_eur_12km_01_mon_198012-208011.nc" library(stars) r = read_ncdf(f, ncsub = cbind(start = c(1, 1, 1, 1), count = c(418, 406, 3, 1)), eps=1e-3) rx...

A similar case was found in #52