A Python library for Meteorology and Climatology.
On Ubuntu, use Anaconda, set up a conda environment, and install the iris package from the conda-forge repository. Then you can import iris inside Python when that environment is active:
$ module add anaconda
$ conda create -n iris -c conda-forge iris
$ conda activate iris
(iris) $ python
>>> import iris
Ubuntu 18.04 instructions were:
$ module add anaconda $ conda create -n iris $ conda activate iris (iris) $ conda install -c conda-forge/label/cf201901 iris (iris) $ python >>> import iris
On Ubuntu 14.04 (Trusty) this is installed alongside the system Python 2.7 so you should just be able to import iris within your scripts.
On Ubuntu 12.04 (Precise) this is provided as a Python virtualenv so you need to load an environment module to use it.
module add python/iris python import iris
This is provided via a Python virtualenv on Precise.
It has so many dependencies that on Trusty I decided to install it in an NFS mounted directory that the system Python includes via sitecustomize.py , and thus take advantage of the Debian packages for many of the dependencies. It still required cartopy, Biggus, and python-netCDF4 installing (also in the NFS directory). This does mean we can't update it easily. If you need to install a newer version on Trusty, make a virtualenv.
Be sure to include the pp_pack file format by doing
module add mo_unpack python setup.py --with-unpack build
The module handles setting the rpath.
For Xenial, I built or installed all of the dependencies from here http://scitools.org.uk/iris/docs/v1.9.2/installing.html including all of the optional ones except for mo_pack. I did however need to build libmo_unpack. Most/all of the others were either in pip or the repos.
Iris dependencies: biggus, pyugrid, cf-units, pyke. All installed with pip install $package --prefix=/usr/local and then deb made.
For building Iris:
git clone https://github.com/SciTools/iris.git
Switch to the latest version, in my case:
git checkout v1.9.2
otherwise it's on the development branch. Then:
python setup.py --with-unpack build python setup.py build pyke_rules python setup.py build python setup.py install --prefix=/usr/local
And finally build the deb.
