Oop Is Shifting Between Domains, Not Disappearing

Why even comment in an article about those topics then?

Using classes hasn't been a part of the definition of OOP since the Treaty of Orlando. Pre-ECMAScript-2015 JS is a mainstream OOP language that doesn't have classes, just prototypes. (Arguably ECMAScript 2015 classes aren't really classes either.)

I see what you did there

I think this is why FP is becoming more popular these days but I'm not sure some people get why. The problem with OOP is you take a data set and spread it all over a 'system' of stateful (mutable) objects and wonder why it doesn't/can't all fit back into place when you need it to. OOP looks great on paper and I love the premise but...

With FP you take a data set and pass it through a pipeline of functions that give back the same dataset or you take a part of that data out, work on it and put it straight back. All your state lives in one place, mutable changes are performed at the edges, not internally somewhere in a mass of 'instances'.

I think micro services et al try to alleviate this by spreading the OO system's instances into silos but that just moves the problems elsewhere.

Agreed. I think objects/classes (C++) should be for software subsystems and not so much for user data. Programs manipulate data, not the other way around - polymorphism and overloading can be bad for performance.

Though OOP is just one step - structured programming works on the same problem.

Even plain objects though, the point isn't the inheritance! The point is to put an interface on the data. Inheritance is sometimes useful because, but there is a reason we keep screaming "prefer composition to inheritance" (even though few listen)

Data hiding is just one of the concepts of OOP. Polymorphism is another one.

How that's implemented is another question. You can do OOP in plain C, several libraries kinda did that, like GTK. Other languages tried to support these concepts with less boilerplate, giving rise to classes and such. But OOP is not about language features, it's fundamentally a way of designing software.

C is capable enough to program in a way that is basically OOP and without using non-idiomatic code. The C++ object system wasn't created in a vacuum.

Not at all OOP is great at simulations, videogames, emergent behaviour in general. If you do crud with oop you will complain about overengineering.

I've heard this a lot in my career. I can agree that most object-oriented languages have had to do a lot of work to make CRUD and database operations easy to do, because they are common needs. ORM libraries are common because mapping between objects and relations (SQL) is a common need.

It doesn't necessarily mean that object-oriented programming is the best for CRUD because ORMs exist. You can find just as many complaints that ORMs obfuscate how database operations really work/think. The reason you need to map from the relational world to the object world is because they are different worlds. SQL is not an object-oriented language and doesn't follow object-oriented ideals. (At least, not out of the box as a standardized language; many practical database systems have object-oriented underpinnings and/or present object-oriented scripting language extensions to SQL.)

> it's super hard without an ORM or at least some kind of object layer

This seems like you might have got caught in something of a tautological loop situation that because you were working in a language with "object layers" it seemed easiest to work in one, and thus work with an ORM.

It might also be confusing the concepts of "data structure" and "object". Which most object-oriented languages generally do, and have good reason to. A good OOP language wants every data structure to be an object.

The functional programming world still makes heavy use of data structures. It's hard to program in any language without data structures. FP CRUD can be as simple as four functions `create`, 'read`, `update`, and `delete`, but still needs some mapping to data structures/data types. That may still sound object-oriented if you are used to thinking of all data structures as "objects". But beyond that, it should still sound relatively "easy" from an FP perspective: CRUD is just functions that take data structures and make database operations or make database operations and return data structures.

A difference between FP and OOP's view of data structures is where "behaviors" live. An object is a data structure with "attached" behaviors which often modify a data structure in place. FP generally relies on functions that take one data structure and return the next data structure. If you aren't using much in the way of class inheritance, if your "objects" out of your ORM have few methods of their own, you may be closer to FP than you think. (The boundary is slippery.)

It's fashionable to dunk on OOP (because most examples - like employee being a subtype of person - are stupid) and ORM (because yes you need to hand write queries of any real complexity).

But there's a reason large projects rely on them. When used properly they are powerful, useful, time-saving and complexity-reducing abstractions.

Code hipsters always push new techniques and disparage the old ones, then eventually realise that there were good reasons for the status quo.

Case in point the arrival of NoSQL and wild uptake of MongoDB and the like last decade. Today people have re-learned the value of the R part of RDBMS.

You don’t want lazy loading. You don’t want to load 1 thing. You don’t want to update 1 thing.

You want to actually exploit RETURNING and not have the transaction fail on a single element in batch.

If you care about performance you do not want ORM at all. You want to load the response buffer and not hydrate objects.

If you ignore ORM you will realize CRUD is easy. You could even batch the actual HTTP requests instead of processing them 1 by 1. Try to do that with a bunch of objects.

I would personally never use ORM or dependency injection (toposort+annotations). Both approaches in my opinion do not solve hard problems and in most cases you don’t even want to have the problems they solve.

That's because you are wrong. There's nothing in relational databases mapping that make objects a better target than even the normal data structures you see in functional languages.

> In my case, I wanted to do CRUD endpoints which were programmatically generated based on database schema.

What is a pure transformation.

The problem is that CRUD applications are an incredibly bad explored area. There only mature projects out there are the OOP ORM ones. That's not because OOP is inherently better suited to the application, it's because there's simply not a lot of people willing to risk into working at that problem.

(And the reason people are not willing can be because developers don't choose their tools through rational evaluation, or may be some irrational one IDK. Mine is certainly because I know if I built an amazing system, nobody would come.)

Microservices are more about making very concrete borders between components with an actual network in between them... and really a contract that has to be negotiated across teams. I feel the best thing this did was force a real conversation around the API boundary and contract, monoliths turn to a big ball of mud once a change slips through that passes in an entire object when just a field is needed, and after a few of these now everything is fairly tightly coupled- modern practices with PRs could prevent a lot of this, but there is still a lot of rubber stamping going on and they don't catch everything. Objects themselves are fine ideas, and I think OOP is great when you focus on composition over inheritance, and bonus points if the objects map cleanly into a relational database schema- once you are starting getting inheritance hierarchies, they often do not. If I had to guess, your experience with OOP is mostly using ORMs where you define the data and it spits out a table for you and some accessor methods, and that works... until it doesn't. At a certain level of complexity the ORM falls apart, and what I have seen in nearly every place I have worked at- is that at some point some innocuous change gets included and now all of a sudden a query does not use an index properly, and it works fine in dev, but then you push it to prod and the DB lights on fire and its really difficult to understand what happened. The style of programming you are talking about would be derided by some old heads as "C with objects" and not "really" OOP. But I do think you are onto something by taking the best parts and avoiding the bad.

"Micro" services aren't great when they are taken to their utmost tiny size, but the idea of a problem domain being well constrained into a deployable unit usually leads to better long term outcomes than a monolith, though its also very true that for under $10k you can get 32 cores of xeons and about 256 gigs of ram, and unless you are building something with intense compute requirements, that is going to get you a VERY long way in terms of concurrent users.

OMG! As an interviewer I would have asked you to elaborate on that response. Perfect opportunity to see if and how the candidate thinks and how deep some pocket of understanding goes.

They are a solution to communication and organizational challenges, not a technical one.

As every other solution they have cons, some of which you have outlined.

The author is complaining about bloat.

The thing is, in this case, the bloat has highly tangible costs: Spreading an application across multiple computers unnecessarily adds both operation costs and development costs.

I rather liked the old post "Object Oriented Programming is an Expensive Disaster that Must End" written over 10 years ago.

https://medium.com/@jacobfriedman/object-oriented-programmin...

Many complained the post was too long, and then debated all kinds of things brought up in the article (such is the way of the internet).

But the one thing I really liked is how it laid out that everyone has a different definition of what OOP is and so it is difficult to talk about.

Why, functions have state -- their closure.

One principal I pursue aggressively in UI design is NOT hiding state!

A client had written a script to calibrate an embedded system across several operating parameters. The program consisted of a TUI to capture setup parameters, then ran 5-layers of nested FOR loop for several hours. This original program lacked even a basic progress bar. You never knew what it was doing or when it would be done - it hid a lot of internal state!

You might say, "the TUI program was hiding relevant state; your program should only hide the irrelevant state." But as I iterated on the program I realized that literally all of the program state was relevant! So I meticulously crafted a GUI to display every operational detail on screen at all times. This included the state of each FOR loop, a plot of the intermediate results, the state of each TCP connection, the state of the unit under test, and the user-selected test parameters, the full path of the calibration files the test relied upon.

While the program had some "magic" (it would auto-recall the default parameters associated with the model under test, and would auto detect equipment types with a network scan) I future-proofed it by ensuring that ALL of the parameters were both visible (not hidden!) and human-overridable AND resettable (A reset-to-default option would appear when the user changed defaults, in case the selection was unintended.) The GUI is also wire silent (no network packets transmitted) until directed by the user to scan for/connect to equipment, and it gives a visual indicator while waiting for network responses.

The GUI also tracked test progress as a first-class file-saveable object, so not only was there a progress bar, but tests could be resumed if equipment connectivity was lost mid-test. And a 95% confidence interval estimate, shown as "between 9 and 13 minutes remaining" to help production techs plan their day.

Even when state is read-only, I still don't hide it. Once a test has commenced, the selectable test parameters cannot be changed, but displaying them on screen confirms to users that their 5-hour test was not started with the wrong parameters.

My big takeaway from this effort is that ANY hidden/inaccessible state is a liability. A user should be able to observe at all times what their program is up to.

OOP is not better suited for user-interfaces than the alternatives.

> Some things are not well-suited to OOP, like linear processing of information in a server. I suspect this is where the FP excitement came from

'Hiding' isn't good enough.

OO sells the dream of 'black-box' objects, where you only need to be concerned with the inputs and outputs, and only the object cares about its internals.

But what do programmers stick inside objects? The world! Files are outside the black-box, but inside a FileWriter object. Likewise with external webservers and databases.

That's where the FP hype comes from - or at least Haskell. I can encapsulate functions perfectly (and actually stop caring about the internals!)

If I try to stuff an Airplane inside flight-recorder, I get a compile error.

Good luck with that if you're a C programmer.

Likewise, I see these patterns as equivalent style choices, since the problem fundamentally dictates the required organization and data flow, because the same optimal solution will be visible to any skilled developer, with these weak style choices of implementation being the only freedom that they actually have.

For example, these two are exactly the same:

    state = concept_operation(state, ...args)

    class Concept:
        def operation(self, ...args)
            self.state = <whatever with self.state>

I suspect people who get emotional about these things haven't spent too much time in the others, to understand why they exist with such widespread use.

It's like food. If you go anywhere and see the common man eating something, there's a reason they're eating it, and that reason is that's it's probably pretty ok, if you just try it. There's a reason they're eating it, and it's not that they're idiots.

This comment i posted in an earlier thread rehashing Inheritance vs. Composition for the gazillionth time is highly relevant here - https://news.ycombinator.com/item?id=45943135

OOD/OOP has not gone away, has not shifted etc. but is alive and well; just packaged under different looking gloss. The meta-principles behind it are fundamental to large scale systems development, namely; Separation-Of-Concerns, Modularization, Reuse and Information-Hiding.

I've seen code from people who got the FP "buzz" and suddenly everything had to be FP. It was a nightmare.

The reality is as it always has been, the best programmers write simple, understandable code, regardless of the paradigm they choose.