This is GraalPy, an implementation of the Python language. A primary goal is to support SciPy and its constituent libraries. GraalPy can usually execute pure Python code faster than CPython (but not when C extensions are involved). GraalPy currently aims to be compatible with Python 3.10, but it is some way from there. While your specific workload may function, any Python program that uses external packages could hit something unsupported. At this point, the Python implementation is made available for experimentation and curious end-users.
The easiest option to try GraalPy is Pyenv, the Python version manager.
It allows you to easily install different GraalPy releases.
To get version 22.3.0, for example, just run pyenv install graalpy-22.3.0.
To try GraalPy with a full GraalVM, including the support for Java embedding and interop with other languages, you can use the bundled releases from www.graalvm.org.
Another option is to use Conda-Forge.
To get an environment with the latest GraalPy, use conda create -c conda-forge -n graalpy graalpy.
- mx - a separate Python tool co-developed for GraalVM development. This tool must be
downloaded and put onto your PATH:
git clone https://github.com/graalvm/mx.git export PATH=$PWD/mx:$PATH - LabsJDK
The following command will download and install JDKs to built GraalVM upon. If successful, it will print the path to set into your JAVA_HOME.
mx fetch-jdkRun mx --dy /compiler python-gvm in the graalpython repository root. If the build is fine, it will print the full
path to the graalpy executable as the last line of output.
For more information and some examples of what you can do with GraalPy, check out the reference.
The best way of using the GraalVM implementation of Python is out of a virtual environment. To do so execute the following command in the project directory:
graalpy -m venv <dir-to-venv>
To activate the environment in your shell session call:
source <dir-to-venv>/bin/activate
In the venv, multiple executables are available, like python, python3 and graalpy.
You should be able to use the pip command from a GraalPy venv to install packages.
Our pip ships some patches for packages that we test internally, these will be applied automatically where necessary.
Support for as many extension modules as possible is a high priority for us.
We are actively building out our support for the Python C API to make extensions such as NumPy, SciPy, Scikit-learn, Pandas, Tensorflow and the like work fully.
This means that some might already work, but we're still actively working on compatibility especially with native extensions.
We have a document describing how we implement the cross-language interop. This will hopefully give you an idea how to use it.
We are working on a mode that is "mostly compatible" with some of Jython's features, minus of course that Jython implements Python 2.7 and we implement Python 3.10+. We describe the current status of the compatibility mode here.
If you're thinking about contributing something to this repository, you will need to sign the Oracle Contributor Agreement for us to able to merge your work. Please also take note of our code of conduct for contributors.
To get you started, we have written a bit about the structure of this interpreter that should show how to fix things or add features.
This GraalVM implementation of Python is copyright (c) 2017, 2019 Oracle and/or its affiliates and is made available to you under the terms the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/. This implementation is in part derived from and contains additional code from 3rd parties, the copyrights and licensing of which is detailed in the LICENSE and THIRD_PARTY_LICENSE files.