It brings to the table a bit of functional programming by reducing amount of state.
It can protect you from stupid errors in rare but painful cases.
It's useful in case of working with kinda global data like singleton. For example you have this:
class SomeSingetonClass
{
// singleton code...
public readonly Data = new Data();
}And you decided to cache Data in a private field inside another class
to stop use Data.FieldName through whole chain like:
LoadPieceOfData(SomeSingetonClass.Instance.Data.FieldName);Especially if you have a few layers of nested objects above Data.
You definitely don't expect Data changed by deserialisation or by somebody in another thread.
And if it happend, you will have two different Data - one cached, one in the singleton.
To prevent that you can add [DoNotCopy] attribute to the field.
class SomeSingetonClass
{
// singleton code...
[DoNotCopy]
public readonly Data = new Data();
}You cannot copy the object anymore, now you have one less state.
It will lead to a compilation error and you'll have to use your changable global data directly.
It's not handy unfortunately but keeps you away from splitting the state. Use it wisely.