Skip to content

Latest commit

 

History

History
94 lines (61 loc) · 2.86 KB

configuration.rst

File metadata and controls

94 lines (61 loc) · 2.86 KB
Raw

Configuring QCoDeS

QCoDeS uses a JSON based configuration system. The configuration is modeled after the module structure plus some higher level switches.

Config files

QCoDeS config files are read in the following order:

  • Package defaults
  • User's home directory
  • QCODES_CONFIG environment variable
  • Current working directory

This means that something specified in your current working directory will overwrite the package defaults.

Config files are backed by a json-schema, meaning that they will be validated before use, and if an error occurs the default settings are kept, and the user is informed.

Note

A change in the configuration requires reimporting the package, or restarting the notebook kernel.

Default config

The following table explains the default configuration of qcodes.

../../qcodes/configuration/qcodesrc_schema.json

Using the config

QCoDeS exposes the default values in the module namespace as config

>>> import qcodes >>> qcodes.config Current values: {'core': {'default_fmt': 'data/{date}/#{counter}_{name}_{time}', 'register_magic': True, 'db_location': '~/experiments.db', 'db_debug': False, 'loglevel': 'WARNING', 'file_loglevel': 'INFO'}} Current paths: ['c:\users\jenielse\source\repos\qcodes\qcodes\configuration\qcodesrc.json', 'C:\Users\jenielse\qcodesrc.json'] <qcodes.configuration.config.Config object at 0x0000029B9D549A90>

Values can be retrieved by key :

>>> qcodes.config['user'] {}

Values can be retrieved by dotted keys:

>>> qcodes.config['gui.notebook'] True

Values can be described:

>>> qcodes.config.describe("gui") "controls gui of qcodes.n Current value: {'notebook': True, 'plotlib': 'matplotlib'}. Type: object. Default: Not defined."

qcodes.config looks like a dictionary (and supports all the dictionary operations you can think about), but it has some additional helpers.

Updating a value

qcodes.config lets you insert a new value which gets stuffed in the "user" part of the config

>>> qcodes.config.add("foo", "bar")

>>> qcodes.config["user.foo"] 'bar'

And also pass a type validator and a description:

>>> qcodes.config.add("foo", "bar", "string", "finds majorana")

>>> qcodes.config.describe("user.foo") 'finds majorana. Current value: bar. Type: string. Default: Not defined.'

Saving

If you made modifications you may want also to save them. F.ex. this

>>> qcodes.config.save_to_home()

Will do what you think, i.e. saving to your home directory. There is the possibility to save to env variable, and current working directory.

More

For a overview of the API, read this: config_api .

Note

for now are supported all the JSON types MINUS enum

add GUI for creating config, explain saving (note on config loaded at module import so no effect if changed at runtime).