redsun is a Python toolkit for building your own modular software for scientific data acquisition. It builds on the Bluesky ecosystem and makes no assumptions about your hardware, so each lab can build the control software its experiments need.
To learn redsun, start with the tutorials, where you build one application step by step without any hardware.
Note
redsun is usable today, but it's still maturing, so expect breaking changes.
When research depends on controlling hardware, one of the hardest problems is making different instruments work together in workflows that are reusable, reliable and documented. On top of that, the software needs an interface that people with less technical background can understand and use correctly.
That's hard because experiments keep changing. It's next to impossible to predict everything the software will finally need to do, especially when the people using it are scientists with no engineering background.
So rather than shipping one complete program, redsun follows the idea of component-based development: it ships ready-made components, and you assemble and wire them for what you need.
In component-based development, what matters most is the interface of each component. A component says what it needs to be built, and offers features that other components can use.
You put components together in a session, which builds the application for you, so you can focus on what each component does.
from collections.abc import Iterator
from typing import Annotated
from mylab.devices import MyMotor
from mylab.presenters import MyController
from mylab.views import MyView
from redsun import AsDevice, AsPresenter, AsView, Declare, Link
from redsun.qt import QtSession
class MyApp(QtSession):
stage: Annotated[AsDevice[MyMotor], Declare(axis=["X", "Y"], egu="mm")]
ctrl: Annotated[AsPresenter[MyController], Declare(timeout=2.0)]
panel: AsView[MyView]
def wire(self) -> Iterator[Link]:
yield self.ctrl.sig_position_changed, self.panel.update_position
MyApp({"session": "my-session"}).run()You declare each component once, with the arguments it needs, and wire says which signal reaches which slot. The session then builds everything in order and connects it.
redsun provides the shared code that connects components, so you can use it to ship a whole application or a single reusable component. Through Python entry points, you can also ship an application as a single YAML configuration file, as long as each package that contributes components includes a redsun.yaml manifest.
The same application then looks like this:
# session.yaml
schema_version: 1.0
frontend: qt
session: my-session
devices:
stage:
plugin_name: mylab
plugin_id: my_motor
axis: ["X", "Y"]
egu: mm
presenters:
ctrl:
plugin_name: mylab
plugin_id: my_controller
timeout: 2.0
views:
panel:
plugin_name: mylab
plugin_id: my_view
wiring:
ctrl.sig_position_changed: panel.update_positionThe session looks up each plugin_id in the manifest that the contributing package ships:
# mylab/redsun.yaml
devices:
my_motor: mylab.devices:MyMotor
presenters:
my_controller: mylab.presenters:MyController
views:
my_view: mylab.views:MyViewLaunch it with:
from redsun import Session
Session.from_config("session.yaml").run()Tip
When you launch a session from a configuration file, the component packages it names (mylab in this example) must be installed in the environment you run it from.
Each redsun session is a Device-View-Presenter (DVP) application. DVP resembles the Model-View-Presenter (MVP) architecture, except that the bottom layer of the application is the Device layer, which uses ophyd-async to talk to the hardware.
This design makes a clear point: redsun is first of all about device control, and tries to do it well.
The documentation covers everything else.
redsun is released under the Apache 2.0 license. See the license for the full text.