README.md

PyMGL Docker

Development container

Build the development container (using source files on host):

docker build -f docker/Dockerfile.ubuntu24.04 -t pymgl-dev-ubuntu24.04 .

Run the container mounted to the host filesystem for source files:

docker run -it -v "$PWD/:/app" pymgl-dev-ubuntu24.04 /bin/bash

Then from within the container, run:

uv pip install -e .[dev,test] -v

# tests require virtual display
Xvfb ${DISPLAY} -screen 0 "1024x768x24" -ac +render -noreset -nolisten tcp  &

export MAPBOX_TOKEN=<your token>

# run C++ test suite
/tmp/build/tests/pymgl_test

# run Python test suite
pytest pymgl/tests

You can also run the tests using xvfb-run to sidestep the need to start Xvfb:

xvfb-run -a --server-args="-screen 0 1024x768x24 -ac +render -noreset" /tmp/build/tests/pymgl_test

To copy rendered PNG files from the running container:

docker cp <containerID>:/tmp /tmp/images

NOTE: it is normal for images rendered in MacOS to be slightly different than Linux, particularly those rendered from vector tiles.

To test for memory leaks using the C++ tests, run:

valgrind --show-leak-kinds=definite --track-origins=yes --log-file=/tmp/valgrind-output /tmp/build/tests/pymgl_test

To run a specific test:

valgrind -s --leak-check=full --track-origins=yes --log-file=/tmp/valgrind-output /tmp/build/tests/pymgl_test --gtest_filter="Wrapper.WidthHeightConstructor"

Then copy to host:

docker cp <containerID>:/tmp/valgrind-output /tmp/

To test for memory leaks using the Python tests, run:

valgrind --show-leak-kinds=definite --log-file=/tmp/valgrind-output python3 -m pytest pymgl/tests -vv --valgrind --valgrind-log=/tmp/valgrind-output > valgrind.log