ObservableCollections is a high performance observable collections(`ObservableList<T>`, `ObservableDictionary<TKey, TValue>`, `ObservableHashSet<T>`, `ObservableQueue<T>`, `ObservableStack<T>`, `ObservableRingBuffer<T>`, `ObservableFixedSizeRingBuffer<T>`) with synchronized views and Observe Extension for [R3](https://github.com/Cysharp/R3).
.NET has [`ObservableCollection<T>`](https://docs.microsoft.com/en-us/dotnet/api/system.collections.objectmodel.observablecollection-1), however it has many lacks of features. It based `INotifyCollectionChanged`, `NotifyCollectionChangedEventHandler` and `NotifyCollectionChangedEventArgs`. There are no generics so everything boxed, allocate memory every time. Also `NotifyCollectionChangedEventArgs` holds all values to `IList` even if it is single value, this also causes allocations. `ObservableCollection<T>` has no Range feature so a lot of wastage occurs when adding multiple values, because it is a single value notification. Also, it is not thread-safe is hard to do linkage with the notifier.
ObservableCollections introduces there generics version, `NotifyCollectionChangedEventHandler<T>` and `NotifyCollectionChangedEventArgs<T>`, it using latest C# features(`in`, `readonly ref struct`, `ReadOnlySpan<T>`). Also, Sort and Reverse will now be notified.
Also, use the interface `IObservableCollection<T>` instead of `INotifyCollectionChanged`. This is guaranteed to be thread-safe and can produce a View that is fully synchronized with the collection.
```csharp
public interface IObservableCollection<T> : IReadOnlyCollection<T>
SynchronizedView helps to separate between Model and View (ViewModel). We will use ObservableCollections as the Model and generate SynchronizedView as the View (ViewModel). This architecture can be applied not only to WPF, but also to Blazor, Unity, etc.
The View retains the transformed values. The transform function is called only once during Add, so costly objects that are linked can also be instantiated. Additionally, it has a feature to dynamically show or hide values using filters.
Observable Collections themselves do not implement `INotifyCollectionChanged`, so they cannot be bound on XAML platforms and the like. However, they can be converted to collections that implement `INotifyCollectionChanged` using `ToNotifyCollectionChanged()`, making them suitable for binding.
ObservableCollections has not just a simple list, there are many more data structures. `ObservableList<T>`, `ObservableDictionary<TKey, TValue>`, `ObservableHashSet<T>`, `ObservableQueue<T>`, `ObservableStack<T>`, `ObservableRingBuffer<T>`, `ObservableFixedSizeRingBuffer<T>`. `RingBuffer`, especially `FixedSizeRingBuffer`, can be achieved with efficient performance when there is rotation (e.g., displaying up to 1000 logs, where old ones are deleted when new ones are added). Of course, the AddRange allows for efficient batch processing of large numbers of additions.
If you want to handle each change event with Rx, you can monitor it with the following method by combining it with [R3](https://github.com/Cysharp/R3):
While it is possible to manually handle the `CollectionChanged` event as shown in the example above, you can also create a `SynchronizedView` as a collection that holds a separate synchronized value.
The view can modify the objects being enumerated by attaching a Filter.
```csharp
var list = new ObservableList<int>();
using var view = list.CreateView(x => x.ToString() + "$");
list.Add(1);
list.Add(20);
list.AddRange(new[] { 30, 31, 32 });
// attach filter
view.AttachFilter(x => x % 2 == 0);
foreach (var v in view)
{
// 20$, 30$, 32$
Console.WriteLine(v);
}
// attach other filter(removed previous filter)
view.AttachFilter(x => x % 2 == 1);
foreach (var v in view)
{
// 1$, 31$
Console.WriteLine(v);
}
// Count shows filtered length
Console.WriteLine(view.Count); // 2
```
The View only allows iteration and Count; it cannot be accessed via an indexer. If indexer access is required, you need to convert it using `ToViewList()`. Additionally, `ToNotifyCollectionChanged()` converts it to a synchronized view that implements `INotifyCollectionChanged`, which is necessary for XAML binding, in addition to providing indexer access.
```csharp
// Queue <-> List Synchronization
var queue = new ObservableQueue<int>();
queue.Enqueue(1);
queue.Enqueue(10);
queue.Enqueue(100);
queue.Enqueue(1000);
queue.Enqueue(10000);
using var view = queue.CreateView(x => x.ToString() + "$");
using var viewList = view.ToViewList();
Console.WriteLine(viewList[2]); // 100$
```
In the case of ObservableList, calls to `Sort` and `Reverse` can also be synchronized with the view.
```csharp
var list = new ObservableList<int> { 1, 301, 20, 50001, 4000 };
using var view = list.CreateView(x => x.ToString() + "$");
view.AttachFilter(x => x % 2 == 0);
foreach (var v in view)
{
// 20$, 4000$
Console.WriteLine(v);
}
// Reverse operations on the list will affect the view
list.Reverse();
foreach (var v in view)
{
// 4000$, 20$
Console.WriteLine(v);
}
// remove filter
view.ResetFilter();
// The reverse operation is also reflected in the values hidden by the filter
Once the R3 extension package is installed, you can subscribe to `ObserveAdd`, `ObserveRemove`, `ObserveReplace`, `ObserveMove`, `ObserveReset`, `ObserveClear`, `ObserveReverse`, `ObserveSort` events as Rx, allowing you to compose events individually.
In the case of Blazor, `StateHasChanged` is called and re-enumeration occurs in response to changes in the collection. It's advisable to use the `CollectionStateChanged` event for this purpose.
Because of data binding in WPF, it is important that the collection is Observable. ObservableCollections high-performance `IObservableCollection<T>` cannot be bind to WPF. Call `ToNotifyCollectionChanged()` to convert it to `INotifyCollectionChanged`. Also, although ObservableCollections and Views are thread-safe, the WPF UI does not support change notifications from different threads. To`ToNotifyCollectionChanged(IColllectionEventDispatcher)` allows multi thread changed.
`SynchronizationContextCollectionEventDispatcher.Current` is default implementation of `IColllectionEventDispatcher`, it is used `SynchronizationContext.Current` for dispatche ui thread. You can create custom `ICollectionEventDispatcher` to use custom dispatcher object. For example use WPF Dispatcher:
```csharp
public class WpfDispatcherCollection(Dispatcher dispatcher) : ICollectionEventDispatcher
{
public void Post(CollectionEventDispatcherEventArgs ev)
`ToNotifyCollectionChanged()` can also be called without going through a View. In this case, it's guaranteed that no filters will be applied, making it faster. If you want to apply filters, please generate a View before calling it. Additionally, `ObservableList` has a variation called `ToNotifyCollectionChangedSlim()`. This option doesn't generate a list for the View and shares the actual data, making it the fastest and most memory-efficient option. However, range operations such as `AddRange`, `InsertRange` and `RemoveRange` are not supported by WPF (or Avalonia), so they will throw runtime exceptions.
Views and ToNotifyCollectionChanged are internally connected by events, so they need to be `Dispose` to release those connections.
Standard Views are readonly. If you want to reflect the results of binding back to the original collection, use `CreateWritableView` to generate an `IWritableSynchronizedView`, and then use `ToWritableNotifyCollectionChanged` to create an `INotifyCollectionChanged` collection from it.
```csharp
public delegate T WritableViewChangedEventHandler<T,TView>(TView newView, T originalValue, ref bool setValue);
public interface IWritableSynchronizedView<T,TView> : ISynchronizedView<T,TView>
`ToWritableNotifyCollectionChanged` accepts a delegate called `WritableViewChangedEventHandler`. `newView` receives the newly bound value. If `setValue` is true, it sets a new value to the original collection, triggering notification propagation. The View is also regenerated. If `T originalValue` is a reference type, you can prevent such propagation by setting `setValue` to `false`.
```csharp
var list = new ObservableList<int>();
var view = list.CreateWritableView(x => x.ToString());
In Unity projects, you can installing `ObservableCollections` with [NugetForUnity](https://github.com/GlitchEnzo/NuGetForUnity). If R3 integration is required, similarly install `ObservableCollections.R3` via NuGetForUnity.
In Unity, ObservableCollections and Views are useful as CollectionManagers, since they need to convert T to Prefab for display. Since View objects are generated only once, it's possible to complement GameObjects tied to the collection.
class ObservableList<T> : IList<T>, IReadOnlyList<T>, IObservableCollection<T>, IReadOnlyObservableList<T>
class ObservableDictionary<TKey,TValue> : IDictionary<TKey,TValue>, IReadOnlyDictionary<TKey,TValue>, IObservableCollection<KeyValuePair<TKey,TValue>> where TKey : notnull
class ObservableHashSet<T> : IReadOnlySet<T>, IReadOnlyCollection<T>, IObservableCollection<T> where T : notnull
class ObservableQueue<T> : IReadOnlyCollection<T>, IObservableCollection<T>
class ObservableStack<T> : IReadOnlyCollection<T>, IObservableCollection<T>
class ObservableRingBuffer<T> : IList<T>, IReadOnlyList<T>, IObservableCollection<T>
class RingBuffer<T> : IList<T>, IReadOnlyList<T>
class ObservableFixedSizeRingBuffer<T> : IList<T>, IReadOnlyList<T>, IObservableCollection<T>
class AlternateIndexList<T> : IEnumerable<T>
```
The `IObservableCollection<T>` is the base interface for all, containing the `CollectionChanged` event and the `CreateView` method.
The `Count` of the View returns the filtered value, but if you need the unfiltered value, use `UnfilteredCount`. Also, normal enumeration returns only `TView`, but if you need `T` or want to enumerate pre-filtered values, you can get them with `Filtered` and `Unfiltered`.
