Installing IRx

Published package

pip install pyirx

The distribution is named pyirx; Python code imports irx.

IRx requires Python 3.10 or newer. Translation uses llvmlite. Native object and executable workflows require an LLVM/Clang-compatible toolchain, while Arrow-backed features additionally require a C++ compiler.

PyArrow and arx-arrowcpp-sources are installed dependencies. IRx uses their Arrow C++ include, source, library, and linker metadata for native runtime builds.

Source checkout

git clone https://github.com/arxlang/arx.git
cd arx
mamba env create --file conda/dev.yaml
conda activate arx
poetry install
makim irx.unittests

Direct RecordBatch Python API

The direct irx.record_batch ctypes API currently loads a standalone shared library. Build it from the checkout before first use:

python -c "from irx.builder.runtime.record_batch import build_record_batch_shared_library; build_record_batch_shared_library()"

IRx programs compiled through the Builder use runtime-feature artifact collection instead of this standalone ctypes setup.

Link modes

IRx emits PIC-compatible objects by default for modern PIE-default linkers. If a downstream manual link still requires non-PIE output, pass the equivalent of clang -no-pie or use Arx’s --link-mode no-pie option.