RestorationBucket class

A RestorationBucket holds pieces of the restoration data that a part of the application needs to restore its state.

For a general overview of how state restoration works in Flutter, see the RestorationManager.

RestorationBuckets are organized in a tree that is rooted in RestorationManager.rootBucket and managed by a RestorationManager. The tree is serializable and must contain all the data an application needs to restore its current state at a later point in time.

A RestorationBucket stores restoration data as key-value pairs. The key is a String representing a restoration ID that identifies a piece of data uniquely within a bucket. The value can be anything that is serializable via the StandardMessageCodec. Furthermore, a RestorationBucket may have child buckets, which are identified within their parent via a unique restoration ID as well.

During state restoration, the data previously stored in the RestorationBucket hierarchy will be made available again to the application to restore it to the state it had when the data was collected. State restoration to a previous state may happen when the app is launched (e.g. after it has been terminated gracefully while running in the background) or after the app has already been running for a while.

Lifecycle

A RestorationBucket is rarely instantiated directly via its constructors. Instead, when an entity wants to store data in or retrieve data from a restoration bucket, it typically obtains a child bucket from a parent by calling claimChild. If no parent is available, RestorationManager.rootBucket may be used as a parent. When claiming a child, the claimer must provide the restoration ID of the child it would like to own. A child bucket with a given restoration ID can at most have one owner. If another owner tries to claim a bucket with the same ID from the same parent, an exception is thrown (see discussion in claimChild). The restoration IDs that a given owner uses to claim a child (and to store data in that child, see below) must be stable across app launches to ensure that after the app restarts the owner can retrieve the same data again that it stored during a previous run.

Per convention, the owner of the bucket has exclusive access to the values stored in the bucket. It can read, add, modify, and remove values via the read, write, and remove methods. In general, the owner should store all the data in the bucket that it needs to restore its current state. If its current state changes, the data in the bucket must be updated. At the same time, the data in the bucket should be kept to a minimum. For example, for data that can be retrieved from other sources (like a database or web service) only enough information (e.g. an ID or resource locator) to re-obtain that data should be stored in the bucket. In addition to managing the data in a bucket, an owner may also make the bucket available to other entities so they can claim child buckets from it via claimChild for their own restoration needs.

The bucket returned by claimChild may either contain state information that the owner had previously (e.g. during a previous run of the application) stored in it or it may be empty. If the bucket contains data, the owner is expected to restore its state with the information previously stored in the bucket. If the bucket is empty, it may initialize itself to default values.

When the data stored in a bucket is no longer needed to restore the application to its current state (e.g. because the owner of the bucket is no longer shown on screen), the bucket must be disposed. This will remove all information stored in the bucket from the app's restoration data and that information will not be available again when the application is restored to this state in the future.