Krzysztof Dyba

The code below works without `{sfarrow}`, but requires `{jsonlite}` to parse JSON with metadata and `{wk}` to convert WKB to WKT. ```r library("wk") library("terra") library("arrow") library("jsonlite") # sample data url...

On Windows, R 4.3.2 includes GDAL 3.7.2, but it doesn't seem to support parquet (driver not compiled). Edit: You are right, it's 3.7.2 now, not 3.6.2.

@kalibera, could you please consider adding this driver to GDAL in R? But maybe this won't be necessary if @paleolimbot completes the `{geoarrow}` package?

I can confirm that this causes the crash. Large raster probably means that needs to be processed in blocks. I tested this example with 8 GB RAM, so maybe you...

@rhijmans, does your example only use PROJ (without GDAL)? Like `sf::sf_project()`?

I think the sieve filter is mainly used to smooth categorical data such as classification and that is [expected](https://gdal.org/programs/gdal_sieve.html): > The input dataset is read as integer data which means...

Maybe sample some points, interpolate them and then combine the rasters? ```r library("terra") set.seed(1) f = system.file("ex/elev.tif", package = "terra") r = rast(f) r[40:50, 30:40] = NA smp = spatSample(r,...

I'm not sure if the sieve filter should be used for numeric (continuous) data? See the example below, the result looks strange: - the square in the middle wasn't filled...

@aloboa, did you try reinstall `terra` to use a newer version of GEOS for compilation?

Maybe instead of implementing a new function it would be easier to use bilinear interpolation from `extract()`? ```r library("terra") set.seed(1) f = system.file("ex/logo.tif", package = "terra") r = rast(f)[[1]] idx...