What is QMI?Main Features

  • Device drivers: With device drivers instruments and devices can be controlled
  • QMI contexts: A QMI context envelops instruments, tasks and other RPC objects and can be connected to from outside from another QMI context
  • Remote Procedure Calls: Custom RPC are used to control objects in QMI contexts
  • QMI signals: QMI context objects can also broadcast and receive signals, which can include e.g. data

'Hello World'

  • A QMI context is created using qmi.start("<context_name>") call
  • Always remember to stop the context with qmi.stop()

Controlling an instrument

  • Instruments can be added into contexts with <instrument_object> = qmi.make_instrument("<name>", <ClassName>, <possible_extra_parameters>)
  • Instrument class description can be seen with help(<instrument_object>)
  • Detailed information about the object and variables can be obtained with help(<instrument_object>)
  • The returned instrument object is an RPC proxy object of the actual class object
  • <instrument>.open() needs to be called first before other RPC methods can be used. Don’t forget to call <instrument>.close() afterwards to ensure stable exit behaviour.

Configuring and logging

  • qmi.conf has a JSON-like structure, but allows comments starting with ‘#’
  • qmi.conf location can be given as an argument to qmi.start by setting QMI_CONFIG environment variable
  • Log levels are set within “logging” keyword section
  • Default name and location of the log file are qmi.log and the user’s home directory
  • These can be changed using “logfile” and “log_dir” options in qmi.conf, but also by setting QMI_HOME environment variable.

Accessing an instrument remotely

  • Instruments to be accessed remotely should be defined in qmi.conf - both on the ‘server’ and ‘client’ side
  • Connect to another QMI context using qmi.context().connect_to_peer("<context_name>", peer_address="<ho.st.i.p:port>")
  • Obtain remote instrument control with qmi.get_instrument("<context_name>.<instrument_name>")

Create a task and a 'service'

  • Also QMI tasks need to be defined in qmi.conf. To enable running the task a a service, parameters “enabled” and “program_module” need to be defined.
  • A task consists of a QMI_Task class and a QMI_TaskRunner class. The latter can be customized to include RPC methods in tasks.
  • Task is started with start(), stopped with stop() and after stopping, the task thread should be “joined” with join() for properly exiting the thread.
  • qmi_proc is an executable created while installing QMI. It can be used to start, stop and checking status of (local) QMI tasks running as background processes.

Open-source vs internal code

  • In Github we have open-source QMI and in Gitlab the internal closed-source QMI
  • Users can make requests via opening new issues a.k.a. tickets in the repository
  • You can and are welcome to contribute too! Just follow the guidelines

Releasing and versioning of QMI code

  • While main version-revision development happens through the main branch, also version-revision-specific development can be done through stable branches
  • User can also use tags as references to code at certain point of time
  • The usual way to install QMI is using Pip and a Pypi package of QMI. But, also releases can be downloaded and used for the installation.