# Maintainer: Rich Li pkgname='python-xarray' pkgver=0.9.6 pkgrel=1 pkgdesc="N-D labeled arrays and datasets in Python" url="http://xarray.pydata.org" arch=('any') license=('Apache') makedepends=('python-setuptools') depends=('python-numpy' 'python-pandas') optdepends=('python-netcdf4: NetCDF4 support' 'python-bottleneck: Faster NaN operations' # 'python-pynio: GRIB and other file format support' # (no Python 3 support yet) 'python-scipy: OPeNDAP support' 'python-matplotlib: Plotting support' 'python-dask: Enable out-of-core parallel computation') checkdepends=('python-pytest') replaces=('python-xray') #source=(https://pypi.python.org/packages/source/x/xarray/xarray-"$pkgver".tar.gz) # I'm using the Github source instead of the PyPI source since the Github source includes the data files used for testing source=("$pkgname-$pkgver.tar.gz::https://github.com/pydata/xarray/archive/v$pkgver.tar.gz") md5sums=('980aa689524338de077d8f889e901327') sha1sums=('f49c7e9f5c2cf42a3c3427815c0bd08dcaefa135') sha256sums=('4b68f2fa5158123f87b158015374e035471dbe00d10e445cb43a5e4ab11045e1') build() { cd "$srcdir/xarray-$pkgver" python setup.py build } # NB: the test suite currently fails due to a recent change in Numpy (from 1.11 # to 1.12), which in turn affected bottleneck. Once bottleneck releases the # next version (current is 1.2.0), then it should be fixed. See # https://github.com/pydata/xarray/issues/1208. # check() { # cd "$srcdir/xarray-$pkgver" # pytest xarray # } package() { cd "$srcdir/xarray-$pkgver" python setup.py install --skip-build --prefix=/usr --root="$pkgdir/" --optimize=1 } # vim:set ts=2 sw=2 et: