Prerequisites

PyDECnet requires Python 3.7 or later.  The Python.org website
indicates that versions 3.7 and earlier are now (as of July 2023) no
longer supported.  The current release of PyDECnet has been tested
with Python 3.7 through 3.12 and with 3.13 beta 2 including the "free
threading" version.  Version 3.8 or later is recommended; some
features are not available with version 3.7.

PyDECnet requires only modules that are part of the standard Python
library.

PyDECnet can also use the following optional modules:

1. python-daemon: if present, the --daemon command line option is
   available.
2. PyYAML: if present, the logging configuration file specified by the
   --log-config command line option can be a YAML file.  If not, only
   JSON configuration files are supported.
3. python-pam: if present, username/password authentication for DECnet
   objects is available.
4. pyserial: if present, the "serial" mode of the DDCMP datalink
   is available.

You can check what optional libraries are installed using this
command:

   pip3 list
   
Installation

Installation from the kit

Kits can be found on the DECnet/Python download page, which is
maintained on the HECnet map server:
http://akdesign.dyndns.org:8080/resources/public/index.html

The simplest installation is the "binary" kit, a "whl" file, which can
be directly installed by the "pip" utility.  You can point it at the
URL of the kit on the download server, or download it first and point
"pip" at the downloaded file.  For example:

pip3 install http://akdesign.dyndns.org:8080/resources/public/decnet-1.0.596-py3-none-any.whl

Python "virtual environment"

Python supports the notion of a "virtual environment" which is a
lightweight way of separating custom configurations from each other
and from the regular system installation.  With recent versions of
Python (e.g., 3.12) private packages cannot be installed in the Python
provided as part of the operating system or applications packaging
(Linux distributions, or Mac OS packaging systems like Homebrew).
Instead, these require any optional library modules to be installed
from packager-supplied kits, wich of course is not an option for
PyDECnet.

The supported and recommended way to handle this situation is to
create a "virtual environment" from the Python version to be used, and
then install PyDECnet as well as any desired optional modules in that
virtual environment.  For example:

python3.12 -m venv /usr/local/decnet-venv
/usr/local/decnet-venv/bin/pip install python-daemon python-pam
/usr/local/decnet-venv/bin/pip install http://akdesign.dyndns.org:8080/resources/public/decnet-1.0.596-py3-none-any.whl

PyDECnet can then be started by executing /usr/local/decnet-venv/bin/pydecnet.

Installation from the Github repository

You can get the current code using the Git tool, from the
pydecnet repository on Github:

   git clone https://github.com/pkoning2/pydecnet.git

You can then install from the resulting working directory:

   cd pydecnet
   python3 setup.py install

Note: if you have done this before, depending on tool versions you may
end up with a broken "pydecnet" command, one that reports "Entry point
... not found" at startup.  If so, remove the decnet.egg-info
directory and repeat the operation.

If you still have problems, you can create a source kit and install
from that:

     python3 setup.py sdist
     pip3 install dist/decnet-1.0.tar.gz

Running directly from the Git working directory

If you want to run the current code in the working directory, without
having to go through a reinstall each time there is a change, you need
this command:

     python3 setup develop

That will create the "pydecnet" script in your default binary
directory (typically /usr/local/bin on Linux systems, a suitable
/Library/Frameworks directory on Mac OS) and a link to the Python
modules in the Git working directory.  You can then invoke the
"pydecnet" command to start the current code.

Unit test

PyDECnet comes with a unit test collection.  This is mostly for
development use, but feel free to use it.  You will need the full
sources (source kit or Git working directory), not the binary kit.
After unpacking the sources, the tests are invoked this way:

   cd pydecnet
   python3 -m unittest 

where "pydecnet" is the Git working directory or the top level
directory of the unpacked source tar archive, according to which you
used to get PyDECnet.  You can specify a particular Python 3.x to run
the unit test collection against that particular Python release.  The
optional modules are optional for test as well, but tests that
exercise one of these will be skipped if the necessary optional
component is not installed.

The PyDECnet unit tests are built using the Python standard library
module "unittest".  For more information on how that works, including
other commands to invoke it to run selected tests or get more verbose
output, refer to the Python library manual.

A full unit test run takes about 2 minutes on a fast machine.  

System exerciser

Included in the unit test directory but not run as part of the unit
test suite described above is a test that creates a simulated network
of 15 PyDECnet nodes of various types and with various connections
among them.  It then starts up that entire network and runs a variety
of tests.  At the moment they are: (1) read all the HTTP status pages,
(2) run data tests among 20 randomly chosen pairs of those 15 nodes.
This delivers a basic check of operation of whole PyDECnet nodes, as
opposed to the unit tests which check just one layer or code component
at a time in isolation.

To run the exerciser, simply pass the script to Python.  You may use a
specific version of Python 3 if desired, but Python 3.8 is required.
The "aiohttp" package is recommended; without it the HTTP page test is
skipped. 

   cd pydecnet
   python3 tests/systemtest.py
   
A system exerciser run takes about 2 minutes.