Installation
Requirements
Python >= 3.9 is required. Additionally the following packages are required in order to work properly:
numpy 1.22 or later
Installing numina
The preferred methods of installation are pip or conda, using prebuilt packages.
Using pip
Run:
pip install numina
Pip will download all the required dependencies and a precompiled versión of numina (if it exists for your platform) from PyPI.
Note
If possible, pip will install a precompiled version of numina in wheel format.
If such a version does not exist, pip will download and compile the source code.
Numina has some portions of C and C++ code. You will need a C/C++ compiler
such as gcc or clang (see Building from source below)
Pip can install packages in different locations. You can use the --user option
to install packages in your home directory.
Our recommended option is to perform isolated installations using virtualenv or venv. See Numina Deployment with Virtualenv for details.
Warning
Do not use sudo pip unless you really really know what you are doing.
Using conda
Numina packages for conda are provided in the conda-forge channel. To install the latest version of numina run:
conda update -c conda-forge numina
See Numina Deployment with Conda for details.
Building from source
You may end up building from source if there is not a stable precompiled version of numina for your platform or if you are doing development based on numina.
Prerequisites
You will need a compiler suite and the development headers of Python and Numpy.
If you are building the development version of numina, you will also need Cython to translate Cython code into C/C++.
The development version can can be checked out with:
git clone https://github.com/guaix-ucm/numina.git
Building and installing
To build and install numina, run:
pip install .
You can all install the package in “editable” mode, including the “-e” option:
pip install -e .