Skip to content

Latest commit

 

History

History
285 lines (204 loc) · 9.82 KB

CONTRIBUTING.md

File metadata and controls

285 lines (204 loc) · 9.82 KB
Raw

Contribute to cuCIM

If you are interested in contributing to cuCIM, your contributions will fall into three categories:

  1. You want to report a bug, feature request, or documentation issue
    • File an issue describing what you encountered or what you want to see changed.
    • The RAPIDS team will evaluate the issues and triage them, scheduling them for a release. If you believe the issue needs priority attention comment on the issue to notify the team.
  2. You want to propose a new Feature and implement it
    • Post about your intended feature, and we shall discuss the design and implementation.
    • Once we agree that the plan looks good, go ahead and implement it, using the code contributions guide below.
  3. You want to implement a feature or bug-fix for an outstanding issue
    • Follow the code contributions guide below.
    • If you need more context on a particular issue, please ask and we shall provide.

Code contributions

Your first issue

  1. Read the project's README.md to learn how to setup the development environment
  2. Find an issue to work on. The best way is to look for the good first issue or help wanted labels
  3. Comment on the issue saying you are going to work on it
  4. Code! Make sure to update unit tests!
  5. When done, create your pull request
  6. Verify that CI passes all status checks. Fix if needed
  7. Wait for other developers to review your code and update code as needed
  8. Once reviewed and approved, a RAPIDS developer will merge your pull request

Remember, if you are unsure about anything, don't hesitate to comment on issues and ask for clarifications!

Seasoned developers

Once you have gotten your feet wet and are more comfortable with the code, you can look at the prioritized issues of our next release in our project boards.

Pro Tip: Always look at the release board with the highest number for issues to work on. This is where RAPIDS developers also focus their efforts.

Look at the unassigned issues, and find an issue you are comfortable with contributing to. Start with Step 3 from above, commenting on the issue to let others know you are working on it. If you have any questions related to the implementation of the issue, ask them in the issue instead of the PR.

Setting Up Your Build Environment

The following instructions are for developers and contributors to cuCIM OSS development. These instructions are tested on Linux Ubuntu 16.04 & 18.04. Use these instructions to build cuCIM from the source and contribute to its development. Other operating systems may be compatible, but are not currently tested.

Code Formatting

Python

cuCIM uses isort, and flake8 to ensure a consistent code format throughout the project. isort, and flake8 can be installed with conda or pip:

conda install isort flake8
pip install isort flake8

These tools are used to auto-format the Python code in the repository. Additionally, there is a CI check in place to enforce that the committed code follows our standards. You can use the tools to automatically format your python code by running:

isort --atomic python/**/*.py

Get libcucim Dependencies

Compiler requirements:

  • gcc version 9.0+
  • nvcc version 11.0+
  • cmake version 3.18.0+

CUDA/GPU requirements:

  • CUDA 11.0+
  • NVIDIA driver 450.36+
  • Pascal architecture or better

You can obtain CUDA from https://developer.nvidia.com/cuda-downloads.

Building and Testing cuCIM from Source

First, please clone cuCIM's repository

CUCIM_HOME=$(pwd)/cucim
git clone https://github.com/rapidsai/cucim.git $CUCIM_HOME
cd $CUCIM_HOME

Local Development using Conda Environment (for gcc 9.x and nvcc 11.0.x)

Conda can be used to setup GCC 9.x, CUDA Toolkit (including nvcc) 11.0.x, and other dependent libraries (as shown in ./conda/environments/env.yml) for building cuCIM.

Otherwise, you may need to install dependencies (such as zlib, xz, yasm) through your OS's package manager (apt, yum, and so on).

Creating the Conda Development Environment cucim

Note that ./conda/environments/env.yml is currently set to use gcc 9.x (gxx_linux-64) and CUDA 11.0.x (cudatoolkit & cudatoolkit-dev). If you want to change the version of gcc or CUDA toolkit package, please update ./conda/environments/env.yml before executing the following commands.

conda env create -f ./conda/environments/env.yml
# activate the environment
conda activate cucim

Building libcucim and install cucim (python bindings):

Building libcucim

# `CC` and `CXX` environment variable will be set by default by `gxx_linux-64` package.
# Check if the environments are correctly set.
[ ${CC##$CONDA_PREFIX/bin} = "$CC" ] && >&2 echo "Environment variable CC doesn't start with '$CONDA_PREFIX/bin'"
[ ${CXX##$CONDA_PREFIX/bin} = "$CXX" ] && >&2 echo "Environment variable CXX doesn't start with '$CONDA_PREFIX/bin'"

# set to use nvcc in the conda environment
export CUDACXX=$CONDA_PREFIX/pkgs/cuda-toolkit/bin/nvcc

# build all with `release` binaries (you can change it to `debug` or `rel-debug`)
./run build_local all release $CONDA_PREFIX

The build command will create the following files:

  • ./install/lib/libcucim*
  • ./python/install/lib/_cucim.cpython-*-x86_64-linux-gnu.so
  • ./cpp/plugins/cucim.kit.cuslide/install/lib/cucim.kit.cuslide@*.so

And, it will copy the built library files to python/cucim/src/cucim/clara/ folder:

  • libcucim.so.*
  • _cucim.cpython-*-x86_64-linux-gnu.so
  • cucim.kit.cuslide@*.so

Building cucim(python bindings)

python -m pip install python/cucim

For contributors interested in working on the Python code from an in-place (editable) installation, replace the last line above with

python -m pip install --editable python/cucim

Cleaning build files

You can execute the following command whenever C++ code is changed during the development:

./run build_local all release $CONDA_PREFIX

Once it is built, the subsequent build doesn't take much time.

However, if a build option or dependent packages are updated, the build can be failed (due to CMakeCache.txt or existing build files). In that case, you can remove use the following commands to remove CMakeCache.txt or build folder, then build it again.

  1. Remove CMakeCache.txt for libcucim, cuslide plugin, and the python wrapper (pybind11).
# this command wouldn't remove already downloaded dependency so faster than `clean` subcommand
./run build_local clean_cache
  1. Remove build-* and install folder for libcucim, cuslide plugin, and the python wrapper (pybind11).
# this command is for clean build
./run build_local clean

Building a Conda package

Setup

You can build a conda package on top of cucim Conda environment created by instructions above:

conda activate cucim

First, please make sure that you have conda-build installed:

# Install conda-build if `conda build` command is not available.
! conda build --help > /dev/null && conda install -c conda-forge conda-build

Export necessary environment variables:

export CUDA="$(conda list | grep cudatoolkit-dev | egrep -o "[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+")"
export PYTHON_VER="$(python -c "import sys; print('.'.join(map(str, sys.version_info[:2])))")"
echo "CUDA       : ${CUDA}"
echo "PYTHON_VER : ${PYTHON_VER}"

Then, create conda-bld folder:

CONDA_BLD_DIR=$(pwd)/conda-bld
mkdir -p $CONDA_BLD_DIR

Build

conda build -c conda-forge \
    --dirty \
    --no-remove-work-dir \
    --no-build-id \
    --croot ${CONDA_BLD_DIR} \
    --use-local \
    conda/recipes/libcucim \
    conda/recipes/cucim

# Conda Package files would be available at `conda-bld/linux-64`
ls conda-bld/linux-64/*cucim*

Install

conda install -y -c ${CONDA_BLD_DIR} -c conda-forge \
    libcucim \
    cucim

Building a package (for distribution. Including a wheel package for pip)

Build

You can execute the following command to build a wheel file for pip.

./run build_package

The command would use ./temp folder as a local build folder and build a distribution package into dist folder using dockcross's manylinux2014 docker image.

./run build_package will reuse local ./temp folder to reduce the build time.

If C++ code or dependent packages are updated so the build is failing somehow, please retry it after deleting the temp folder under the repository root.

Install

python -m pip install dist/cucim*.whl

Running Tests

Once cuCIM is installed, you can test the module through ./run test command.

# Arguments:
#   $1 - subcommand [all|python|c++] (default: all)
#   $2 - test_type [all|unit|integration|system|performance] (default: all)
#   $3 - test_component [all|clara|skimage] (default: all)

./run test                      # execute all tests
./run test python               # execute all python tests
./run test python unit          # execute all python unit tests
./run test python unit skimage  # execute all python unit tests in `skimage` module
./run test python unit clara    # execute all python unit tests in `clara` module
./run test python performance   # execute all python performance tests