Golaborate is an application for integrating laboratory hardware into automated testbeds. Example uses include:

1. Corongraphs operated inside a vacuum chamber remotely

2. Ground support equipment to align imaging spectrometers

3. Automated lifetime testing

Users should begin with this README. Golaborate is divided into two repositories, a suite of device drivers and server applications and clients for those servers available in both python and matlab. The repos are:

• Unified Documentation - Golaborate-docs (you are here)
• server - Golaborate
• clients - Golaborate-clients

This documentation is a work in progress. If you are at JPL, contact Brandon Dube (383D) via email or phone for assistance. If you are outside JPL, please open an issue on this repo and assistance will be provided on an as-time-allows basis.

Precompiled releases of multiserver, andorhttp3, andorhttp2, and dacsrv are available on the Github releases page. When self-compile Multiserver is pure Go. The andorhttp binaries depend on the installation of andor's SDK version 2 (andorhttp2) or 3 (andorhttp3). dacsrv uses Cgo but all C dependencies are vendored.

Installing the Go tooling and compilation are outside the scope of this documentation.

To install the clients, simply

$ python -m pip install /~https://github.com/nasa-jpl/golaborate-clients

In the future the clients will be available from PyPi.

Once multiserver is in hand, a configuration file must be made. We'll set up a Physik Instrument motion system and use the mocking feature for the purposes of the demo. On the command line, run:

$ multiserver mkconf

This will make a multiserver.yml file in the current directory, which is essentially empty:

$ cat multiserver.yml
Addr: :8000
Mock: false
Nodes: []

Addr is where the program will listen for requests Mock is a global flag for whether hardware is simulated or real. Nodes is the list of hardware that will be run through multiserver. We'll add two PI controllers:

$ cat multiserver.yml
Addr: :8000
Mock: true
Nodes:
  - Type: pi
    Endpoint: my-pi-controller1
    Addr: 192.168.100.1

  - Type: pi
    Endpoint: my-pi-controller2
    Addr: 192.168.100.2

The port used must be the default for each vendor and not specified in the config file. Serial: True can be added to any node to use the serial (RS-232 or RS-422) interface if it is supported. Then addr would typically be /dev/ttyusbN.

With a valid config, simply start multiserver:

$ multiserver run

The program can be left running idefinitely. On linux a trailing ampersand can be used to run it in the background, or a tmux or screen session used. The logs are written to stderr, which can be redirected to a file to save them.

import motion

c = motion.Controller('localhost:8000/my-pi-controller1', axes=['1', '2', '3'])

x_axis = getattr(c, '1')
x_axis.enable()
x_axis.home()
print(x_axis.pos())
>>> 0
x_axis.velocity(5) # set
print(x_axis.velocity()) # get
>>> 5
x_axis.move_abs(20)
x_axis.move_rel(-5)
print(x_axis.pos())
>>> 15.000000012101383

The axes being labeled 1, 2, 3 is a PI semantic; other motion vendors have different axis naming schemes.

There is no state within the client, an arbitrary number of clients can come and go without doing anything to the server. Similarly, the client can be initialized before the server is turned on. Typically, the server is run as part of the infrastructure for a testbed, and users have no concern for starting or stopping it.

before the colon a specific address (such as localhost) can be used to limit who the server listens for. A trailing slash can be used to modify the rool URL, e.g. Addr: localhost:8000/mytestbed will place hardware at localhost:8000/mytestbed/<endpoint>.