Caching

There are two storage mechanisms for raw data: the database (DB) and context-local storage. The cache is used to signal when the local cache needs to be updated.

During the first run, a checksum is calculated for all registered variables. This checksum acts as a cache key that signals when a value has been updated and needs to be refreshed from the DB.

Values from the DB are retrieved only when at least one setting in the context is requested. In this case, data is fetched from the DB only when required.

Once at least one setting is requested, all raw values are fetched and saved in the context-local storage. This ensures that all settings remain consistent.

Raw data for a setting is converted to a Python object only when the setting is requested. This avoids unnecessary processing of all raw data. Since raw data can be converted to a Python object at any time, this approach is efficient.

For every request, the system checks the cache to verify if the checksum value has changed. If it has, the context-local data is marked as unpopulated.

When the system repopulates (i.e., a setting is requested again after being marked unpopulated), it updates all raw values. However, Python objects are invalidated only if the corresponding raw value has changed.

The cache trigger class is responsible for updating Python objects when the corresponding database values are updated.

The default trigger class is content_settings.cache_triggers.VersionChecksum, and its name is stored in CONTENT_SETTINGS_CACHE_TRIGGER.

Raw to Python to Value

The journey from a database raw value to the setting's final returned value (settings attribute) consists of two stages:

  1. Creating a Python Object from the Raw Value:

  2. Retrieving an updated value involves converting the raw value into a Python object.

  3. Using the give Function:

  4. The give function of the type class converts the Python object into the attribute's final value.

Behavior for Different Types:

  • Simple Types:

  • The give function directly returns the Python object.

  • Complex Types:

  • The give function can:
    • Utilize the context in which the attribute is used.
    • Apply suffixes of the attribute to modify the returned value.

When Is the Checksum Validated?

The checksum is validated in the following scenarios:

  • At the Beginning of a Request:

  • Triggered by signals.check_update_for_request.

  • Before a Celery Task (if Celery is available):

  • Triggered by signals.check_update_for_celery.

  • Before a Huey Task (if Huey is available):

  • Triggered by signals.check_update_for_huey.

Precached Python Values

Experimental Feature

The system allows all Python objects to be precached for each thread or async context at the time it starts (e.g., when the DB connection is initialized).

To activate this feature, set: CONTENT_SETTINGS_PRECACHED_PY_VALUES = True

Note: This feature might cause issues if a new thread or async context is started for every request.

See also content_settings.caching.

Stand With Ukraine