LoginSign Up
← Back to all sparks
G

geotargets

ANALYTICS
Velocity0.0

Targets extensions for geospatial data. Contribute to ropensci/geotargets development by creating an account on GitHub.

github.com/ropensci/geotargets ↗

Geospatial targets grew from two raster helpers into a tiling and multi-backend pipeline layer.

geospatialpipelinesr-packageropensciraster
◆Current state
geotargets extends the targets pipeline framework with target factories that know how to serialise geospatial objects — terra rasters and vectors, stars arrays, raster collections, and VRT references. It completed rOpenSci review and transferred ownership during 0.3.0. Writing behaviour is now configurable through per-target arguments and package-level options.
◆Where it's heading
The arc runs from 'targets can hold a SpatRaster' to 'targets can hold a tiled, dynamically branched raster workflow with controlled datatype and driver.' Recent work is about giving users control over how objects hit disk — datatype, driver, metadata sidecars, pass-through arguments to the underlying writers — which is where correctness problems in geospatial pipelines actually live. External contributors are driving a visible share of it.
◆Prediction
Expect continued work on write-path fidelity and format coverage rather than new target types, since the last two releases both resolved metadata and driver defaults that were silently losing information.

◆Recent moves

  1. 1y ago

    Guards sozip metadata option against GDAL below 3.7

    A CRAN-driven fix that errors clearly instead of failing obscurely when GDAL is too old for the sozip metadata path added in 0.3.0. Housekeeping on the previous release.

    View source ↗
  2. 1y ago

    VRT targets, datatype control, and GPKG default for vectors

    The write path gets serious attention: tar_terra_vrt() references multiple sources through a GDAL virtual dataset, targets gain datatype and type arguments, and the vector default switches to GPKG because shapefiles were dropping CRS information. This is the release where geotargets started caring about what survives a round trip to disk.

    View source ↗
  3. 1y ago

    stars backend and dynamically branched raster tiles

    ⚡ SPARK

    The release that turns geotargets from a serialisation helper into a pipeline design tool — tiles with dynamic branching plus a second geospatial backend, which is the capability everything in 0.3.x refines.

    View source ↗
  4. 2y ago

    First release: raster, vector, and collection targets

    The founding release establishes the pattern the package still follows — a tar_* factory per geospatial class, plus package-level options. Narrow compared with what tiling later made possible.

    View source ↗