CODEWITHSUNDEEP
c#.netimmutability
Miguel
Miguel

Reputation: 3506

Still Another Question About Immutable Objects

Good morning, afternoon or night,

Supposing I have a class

public class SomeClass
{
    uint   SomeField1;
    uint[] SomeField2;

    (...)
}

and a series of immutable objects A1, A2, ... of type SomeClass differing, for example, only by SomeField1, is it a good practice to set SomeField2 in all of them pointing to the same place in memory, or should I use array duplication in the constructors even if I know the objects are immutable?

Thank you very much.

Upvotes: 2

Views: 241

Answers (4)

Eric Lippert
Eric Lippert

Reputation: 660108

If you can guarantee that no consumer of your object ever mutates the array then you can safely re-use the array as you describe. However, that seems like playing with fire to me. In this case I would be strongly inclined to actually make the field a ReadOnlyCollection<uint> that wraps the array; that way you get (1) nice clear "in code" documentation that this collection is read-only, and (2) an exception if someone attempts to violate the read-only-ness.

Upvotes: 1

CodesInChaos
CodesInChaos

Reputation: 108800

In your old question Is this a good practice of immutability? I already stated:

Since you know the content of the other array will never change you can directly use it. You could even share one instance of the array between several instances of your class if you want to without causing any problems

So as long as you can guarantee immutability sharing instances is no problem.

Upvotes: 1

BrokenGlass
BrokenGlass

Reputation: 160902

"The man" Eric Lippert has a great blog post on this which covers shallow vs deep immutability and observational immutability, this seems to be at the core of your question. If you used a reference to an array in your constructor and didn't create your own copy of the array you would only have "shallow" immutability.

The ints field is “shallowly” immutable. You can rely upon it being immutable to a certain extent, but once you reach a point where there is a reference to a mutable object, all bets are off.

Obviously the opposite of shallow immutability is “deep” immutability; in a deeply immutable object it is immutable all the way down.

In your case SomeField2 would also have observational immutability: From the point of view of any caller the field would be immutable (provided you don't grant direct access to it as @Slaks pointed out) - unless someone else changes the array you reference. If you are guaranteed that that array never changes that's all you need - this would be one application of the flyweight pattern.

Upvotes: 3

Gabe
Gabe

Reputation: 86718

One of the main reasons to use immutable data structures is that they can safely share data. If your array is never exposed and it can never change, then by all means, you should be sharing it amongst other immutable objects that need the same array.

Upvotes: 1

Related Questions