Skip to content

Persistence

RoSys' PersistentModule provides an easy interface to backup and restore parts of the object state. The following example demonstrates a Model class that has value, which is manipulated with a ui.slider.

#!/usr/bin/env python3
from typing import Any

from nicegui import ui

from rosys import persistence


class Model(persistence.PersistentModule):

    def __init__(self) -> None:
        super().__init__()
        self.value: float = 1.0

    def restore(self, data: dict[str, Any]) -> None:
        self.value = data.get('value', 1.0)

    def backup(self) -> dict[str, Any]:
        return {
            'value': self.value,
        }


model = Model()
ui.slider(min=0, max=10.0, step=0.1).bind_value(model, 'value').props('label-always')

ui.run(title='RoSys')

By deriving from PersistentModule and implementing backup and restore, RoSys will automatically write the value to a file in the directory ~/.rosys/. The filename contains the name of the module. After restarting the script, the value will be restored to its last state.

The request_backup method can be called to enforce a backup within RoSys' next backup cycle, which happens every 10 seconds. During shutdown, all backups are performed, independent of whether request_backup has been called.

The backup function can return any JSON-serializable dictionary that represents the current state. It should match the restore function so that it can translate it back to object state.

You should choose wisely which values to persist. Try to avoid consuming unnecessary CPU and IO bandwidth for volatile things like wheel odometry or other sensor readings.

Note that the persistence module contains a number of helper functions:

  • to_dict: converts (dictionaries or lists of) dataclasses into a dictionary (or list)
  • from_dict: converts a dictionary into a dataclass of given type
  • replace_dict: replaces the content of a dictionary using from_dict for each item
  • replace_list: replaces the content of a list using from_dict for each item
  • replace_set: replaces the content of a set using from_dict for each item
  • replace_dataclass: replaces the attributes of a dataclass with the values of a dictionary