Development¶

APyTypes is an open source project and contributions are very welcome! From bug reports, through enhancement suggestions, to better documentation and code.

The development happens at https://github.com/apytypes/apytypes where you also can provide suggestions and report bugs.

Here, there will be some development information to be expanded over time.

Pre-requisites¶

Most required packages will be installed during the installation process, but a few things are required

A C/C++ compiler supporting at least C++17.
CMake

Clone the repository¶

To clone the repository, do

git clone https://github.com/apytypes/apytypes.git

However, if you want to contribute to the development, it is preferred to first fork your own copy and then clone that. See the next section.

Fork and clone¶

Go to the GitHub-repository and then click the Fork button. This will create a copy of APyTypes in your own user space. Now, click the Code button and clone using either ssh (preferred) or https. For ssh to work you need to set up ssh keys.

Installation¶

To install APyTypes from a cloned git-version, make sure that you stand in the directory containing the cloned git-version, then do

python -m pip install .

Any change in the C++-code will require a reinstallation, so this is something done frequently during development.

If you want to run tests, instead doing

python -m pip install .[test]

will also install the test dependencies required. (This only needs to be done once, or when the dependencies are updated.)

Build using 32-bit limbs on 64-bit platforms¶

If you want to build APyTypes with 32-bit limbs on a 64-bit platform, do

CXXFLAGS="-DCOMPILER_LIMB_SIZE=32" python -m pip install -v .

This can reduce the memory requirements when using short fixed-point numbers and also speed up some computations as more computations can be done in a SIMD-instruction, while slowing down others as long wordlength operations will be split over more limbs.

Run test suite¶

To run all test cases do

python -m pytest lib

Update stub files¶

There is a script in the root directory stubgen.sh that must be executed to update the stub-files in case any additional Python-level methods or classes are added or if argument names or Python-level docstrings are changed.

Run floating-point tests vs Berkeley TestFloat¶

Randomly generated test cases from Berkeley TestFloat can be run for the floating-point datatypes. These are not executed by using pytest but instead with a dedicated Python script. To use it, start by installing Berkeley TestFloat, and make sure to add the binary testfloat_gen to your PATH. You can then see example usage of the script by running

cd lib/test/apyfloat/
python run_berkeley_cases.py -h

These are intended to more long running tests. If any bugs or deviations are found, dedicated tests should be added to the regular test suite.