Install this version:
emerge -a =dev-python/numba-0.66.0-r1
If this version is masked, you can unmask it using the autounmask tool or standard emerge options:
autounmask =dev-python/numba-0.66.0-r1
Or alternatively:
emerge --autounmask-write -a =dev-python/numba-0.66.0-r1
# Copyright 1999-2026 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2
EAPI=8
DISTUTILS_USE_PEP517=setuptools
PYTHON_COMPAT=( python3_{12..14} )
inherit distutils-r1 optfeature
DESCRIPTION="NumPy aware dynamic Python compiler using LLVM"
HOMEPAGE="https://numba.pydata.org/
https://github.com/numba"
SRC_URI="https://github.com/numba/numba/archive/${PV}.tar.gz -> ${P}.gh.tar.gz"
LICENSE="BSD"
SLOT="0"
KEYWORDS="~amd64 ~arm64"
IUSE="openmp threads"
RDEPEND="
>=dev-python/llvmlite-0.48.0[${PYTHON_USEDEP}]
<dev-python/llvmlite-0.49[${PYTHON_USEDEP}]
>=dev-python/numpy-1.22[${PYTHON_USEDEP}]
<dev-python/numpy-2.5[${PYTHON_USEDEP}]
threads? ( >=dev-cpp/tbb-2019.5 )
"
# numpy is needed to build, not only to run, and at a higher floor than at
# runtime. setup.py get_ext_modules() does `import numpy` for np.get_include()
# to compile the C extensions, and sets min_numpy_build_version = 2.0.0rc1
# against min_numpy_run_version = 1.22 -- so RDEPEND above is right for running
# and wrong for building. 0.66.0 ships no pyproject.toml, so the eclass
# synthesises one and nothing pulls numpy in for the build phase.
# verified 2026-07-27
BDEPEND="
dev-python/pip[${PYTHON_USEDEP}]
dev-python/versioneer[${PYTHON_USEDEP}]
>=dev-python/numpy-2.0.0[${PYTHON_USEDEP}]
<dev-python/numpy-2.5[${PYTHON_USEDEP}]
"
pkg_setup() {
if ! use openmp; then
export NUMBA_DISABLE_OPENMP=1 || die
else
unset NUMBA_DISABLE_OPENMP || die
fi
if ! use threads; then
export NUMBA_DISABLE_TBB=1 || die
else
unset NUMBA_DISABLE_TBB || die
export TBBROOT="${EPREFIX}/usr" || die
fi
}
python_compile() {
# FIXME: parallel python building fails. See Portage bug #614464 and
# gentoo/sci issue #1080.
export MAKEOPTS=-j1 || die
distutils-r1_python_compile
}
pkg_postinst() {
optfeature "compile cuda code" dev-util/nvidia-cuda-toolkit
}
Manage flags for this package:
euse -i <flag> -p dev-python/numba |
euse -E <flag> -p dev-python/numba |
euse -D <flag> -p dev-python/numba
>=dev-python/llvmlite-0.48.0[] <dev-python/llvmlite-0.49[] >=dev-python/numpy-1.22[] <dev-python/numpy-2.5[] threads? ( >=dev-cpp/tbb-2019.5 )
dev-python/pip[] dev-python/versioneer[] >=dev-python/numpy-2.0.0[] <dev-python/numpy-2.5[]