Skip to contents

Getting Started

Source: vignettes/getting-started.qmd

What the package installs

gdalraster.windows downloads a verified GDAL build runtime, compiles gdalraster against it, and vendors the complete native dependency closure into the installed gdalraster package. The result carries:

  • GDAL and dependency DLLs under libs/x64;
  • matching GDAL and PROJ data under gdal/ and proj/;
  • pure-Python osgeo_utils under python/; and
  • a provenance manifest recording the bundle used for the build.

No .Rprofile hook, PATH edit, DLL preload, or GDAL/PROJ environment variable is needed in later sessions.

Prerequisites

  • Windows and R.
  • The matching Rtools release. Compilation is the only step the prebuilt GDAL bundle cannot remove.

One-time setup

Install the package, then build into your regular user library:

pak::pak("jimbrig/gdalraster.windows")
gdalraster.windows::gdal_setup()

gdal_setup() prints one plan, installs or skips the runtime idempotently, builds and vendors gdalraster, provisions embedded Python when a system CPython is available, and verifies the result in fresh processes.

Interactive sessions ask before replacing an existing gdalraster in the destination library; non-interactive replacement requires force = TRUE.

For a non-destructive isolated library:

result <- gdalraster.windows::gdal_setup(isolated = TRUE)
result$lib

Everyday use

After setup, gdalraster.windows does not need to be attached:

For an isolated install, load from that library:

library(
  gdalraster,
  lib.loc = file.path(
    tools::R_user_dir("gdalraster.windows", "data"),
    "library"
  )
)

Status and verification 

gdalraster.windows::gdal_sitrep()
gdalraster.windows::gdal_verify()

gdal_verify() uses fresh Rscript --vanilla processes. It checks the Algorithm API, GEOS and CRS support, Arrow, Parquet, HDF5, and netCDF driver registration, a first Parquet open, and the GeoPackage Python validator when Python provisioning is ready.

Updating 

gdalraster.windows::gdal_update()

The runtime and built package manifests let this operation distinguish a current build from one that must be rebuilt.

Offline setup

Download a bundle release asset on a connected machine, transfer it, then use:

gdalraster.windows::gdal_setup(
  local_zip = "C:/Downloads/gdal-bundle-v3.13.2-windows-x64.zip"
)

See the Runtime Guide for layout and provenance details, Architecture for the loading model, and Troubleshooting for diagnosis.