Setting up a development environment

Build requirements

See Build requirements


Vagrant is a tool that works with a virtualization product such as VirtualBox to create a reproducible development environment. GDAL includes a Vagrant configuration file that sets up an Ubuntu virtual machine with a comprehensive set of dependencies.

Once Vagrant has been installed and the GDAL source downloaded, the virtual machine can be set up by running the following from the source root directory:

# VAGRANT_VM_CPU=number_of_cpus
vagrant up

The source root directory is exposed inside the virtual machine at /vagrant, so changes made to GDAL source files on the host are seen inside the VM. To rebuild GDAL after changing source files, you can connect to the VM and re-run the build command:

vagrant ssh
cmake --build .

Note that the following directories on the host will be created (and can be removed if the Vagrant environment is no longer needed):

  • ../apt-cache/ubuntu/jammy64: contains a cache of Ubuntu packages of the VM, to allow faster VM reconstruction

  • build_vagrant: CMake build directory

  • ccache_vagrant: CCache directory


The Linux environments used for building and testing GDAL on GitHub Actions are defined by Docker images that can be pulled to any machine for development. The Docker image used for each build is specified in linux_build.yml. As an example, the following commands can be run from the GDAL source root to build and test GDAL using the clang address sanitizer (ASAN) in the same environment that is used in GitHub Actions:

docker run -it \
    -v $(pwd):/gdal:rw \
cd /gdal
mkdir build-asan
cd build-asan

To avoid built objects being owned by root, it may be desirable to add -u $(id -u):$(id -g) -v /etc/passwd:/etc/passwd to the docker run command above.

Building on Windows with Conda dependencies and Visual Studio

It is less appropriate for Debug builds of GDAL, than other methods, such as using vcpkg.

Install git

Install git

Install miniconda

Install miniconda

Install GDAL dependencies

Start a Conda enabled console and assuming there is a c:\dev directory

cd c:\dev
conda create --name gdal
conda activate gdal
conda install --yes --quiet curl libiconv icu git python=3.7 swig numpy pytest zlib clcache
conda install --yes --quiet -c conda-forge compilers
conda install --yes --quiet -c conda-forge \
    cmake proj geos hdf4 hdf5 \
    libnetcdf openjpeg poppler libtiff libpng xerces-c expat libxml2 kealib json-c \
    cfitsio freexl geotiff jpeg libpq libspatialite libwebp-base pcre postgresql \
    sqlite tiledb zstd charls cryptopp cgal librttopo libkml openssl xz


The compilers package will install vs2017_win-64 (at time of writing) to set the appropriate environment for cmake to pick up. It is also possible to use the vs2019_win-64 package if Visual Studio 2019 is to be used.

Checkout GDAL sources

cd c:\dev
git clone

Build GDAL

From a Conda enabled console

conda activate gdal
cd c:\dev\gdal
cmake --build build --config Release -j 8

Setting development environment variables

Once GDAL has been built, a number of environment variables must be set to be able to execute C++ or Python utilities of the build directory, or run tests.

This can be done by sourcing the following from the build directory:

. ../scripts/

(with adjustments to the above path if the build directory is not a subdirectory of the GDAL source root).

For Windows, a similar scripts/setdevenv.bat script exists (it currently assumes a Release build).

To verify that environment variables have been set correctly, you can check the version of a GDAL binary:

gdalinfo --version
# GDAL 3.7.0dev-5327c149f5-dirty, released 2018/99/99 (debug build)

and the Python bindings:

python3 -c 'from osgeo import gdal; print(gdal.__version__)'
# 3.7.0dev-5327c149f5-dirty