Hacker News Reader

Comment by tsimionescu | Hacker News Reader

The problem this is trying to resolve is creating copies of a complex immutable class with some deep fields changed. I do have to wonder for a case like this if using a mutable object with a deep copy function wouldn't be a better solution, rather than adding extra magic to do it more easily as possible...

Edit: The code in the article would become something like

  pendingOrders.map(order -> { 
    copy = order.DeepCopy();   
    copy.getApproval()
        .getConfirmation()
        .setUpdatedOn(LocalDateTime.now())); 
    return copy;
    } 
  )

(of course, we could also get rid of the getters and setters and replace them with direct field access, if we take the data class concept seriously)

View on HN

The issue is that this potentially copies a lot of unnecessary information. Also, if you depend on object identities (reference equality), deep copying will recreate objects even if they aren't supposed to change. Furthermore, if you want to chain two update functions, you'll deep copy parts of the object twice times.

Lenses are "smart" in the sense that they only copy what's necessary.

A business method on order would do the same without any unnecessary overhead and ensuring proper encapsulation and consistency of the state that cannot be achieved with getters and setters.

   pendingOrders.map(order -> order.approvalConfirmationUpdated(now())

I believe the intention is to allow operations on multilayered business objects to be composable. So your approvalConfirmationUpdated() method may actually be made up of a sequence of smaller operations, each of which I might want to use elsewhere in the application. Lenses allow this to happen succinctly, and works better with eg the Java streams API.

I got that, my point is that such composability is rarely better than good old OOP with encapsulation. Enumerating all possible state transitions on object interface leads to better design than procedural programming (and composable operations are procedural programming).

I don't think I understand. We're talking about how to implement this. That method will still have to use deep copying + mutation, lenses or something else internally.

You do not need lenses for implementation of a method working with internal state, that will be a gross over-engineering.

   record Order(List<Approval> approvals, int version, OrderStatus status) {
     Order confirmed(Instant timestamp, UUID approver) {
         var approvals = this.approvals.stream().map(a-> a.user().uuid().equals(approver) ? a.confirmed(timestamp) : a).toList();
         return new Order(approvals, version++, OrderStatus.APPROVED);
   }

Lenses are an abstraction of what you mentioned. For a single example it's of course overengineering. The benefit is that e.g. when you have a bunch of methods like that, you can avoid duplicating the code that is responsible for copying the inner layers. Otherwise if you e.g. add a layer, you have to touch all those methods.

If you look at my example you will notice that it has zero lines that would be duplicated in real life scenarios, because it does not perform a deep copy (a benefit of using immutable objects).

Imagine you changed it from an array of approvals to a single one like in the original example - you'd need to make a change in every method of that type (replacing the map). That's code duplication, it's just not really obvious yet because there's only two layers to pass through. Per layer you need a constructor call (or map to copy & modify the array).

As a more obvious example, if you want to modify a.b.c.d.e (which isn't unrealistic), you'll need to call the constructors of A, B, C and D. If you don't use lenses, this is the code that will be duplicated. You can spread it between the classes or do all that in a method of A, but if you want to also modify a.b.c.d.f, you'll need to duplicate all that code (add another method to A, B, C, D that each calls the constructor).

With lenses, you define once how to access d from a and then any modification of d can happen through that. If the structure changes, you only need to do the changes once by modifying the lens.

Those things are easy to imagine, however:

1) change in cardinality is such a change in domain that lenses won’t solve it. There will be much bigger changes in business logic probably making original code obsolete regardless of used pattern.

2) deep tree modifications of the kind that you mentioned indicate problems with architecture. Why would you need the whole typed tree (not DOM or something, but object tree) to modify a tiny leaf of it? If you touch multiple leafs with root as closest integration point, why your data model is designed like that, pointing to strong coupling of different contexts?

3) most importantly the code that can be reused, can be extracted to a private method of an entity where the change occurs. Calling nested constructors on the root object breaks encapsulation — it should pass the message to nested objects instead.

If you have an immutable structure, you can't just modify the leaf of it. Unless you're arguing that deep, immutable structures are an issue in themselves?

Oh, we cannot modify it, of course, but we are talking about the ways of producing a modified copy. The part of the argument about deep immutable structures is just a side remark: building them for modification is likely pointing to a problem with design, but that’s not the most important part. The important part is that it is better to call business methods than externalizing the whole thing with deconstruction-modification-construction to a lens or some other code, when you have the full power of OOP to do it right.

This sounds eerily similar to persistent data structures (eg in FP contexts like Clojure): updated values share as much[1] as possible with their inputs, and create new values only[1] when they differ. Granted in FP contexts, this is primarily an optimization, transparent[2] to the programmer actually using those values in a program.

From the perspective of such a programmer, I can’t think of a scenario where I’d want both value and reference equality semantics for the same objects. Is it reasonable to assume that the “smartness” here is likewise focused on making immutability perform well, rather than on use cases where value and reference equality are simultaneous considerations?

1: Handwaves away implementation details. General cautions about abstractions leaking apply.

2: Exceptions may apply[1].

Yeah, it's also originally from the FP side. There it started I believe because people were looking how to easily make immutable updates to deep structures, since the normal way is pretty boilerplate heavy in comparison to the mutation way of just chaining accessors for an update.

And you're close, it's about using immutability to make value equality cheap by making reference equality a proxy for it. If you don't use mutations, value equality implies reference equality and is thus equivalent (since reference equality already normally implies value equality). That means you can get away with just a single pointer comparison in comparison to completely traversing both structures. This is e.g. what React does to determine whether arguments of a component have changed.

(Well, at least I also struggle to think of a scenario where both types of equality are semantically important).

Reply on news.ycombinator.com