I look at OOP Patterns as standards and practices.
The same way we have building codes for staircases the framing of walls and electrical installations to prevent injury or collapse or fire.
Sure, you can dodge a lot of design pattern paradigms and still make a working application that makes money. You can also invent your own system when building your house and maybe nothing bad will happen. That tragedy hasn’t yet struck does not make the building codes bad just because you got away with it.
The *concept* of patterns makes sense. A shared language that developers can use when building things.
The *reality* of patterns has been much less useful. The original ones were indeed a reaction to warts in the popular languages of their era. And as we tend to do in our industry, these have been cargo culted along the way and for some reason I still see people talking about them as first class citizens 30 years later.
People don't seem to realize that patterns should be and are fluid, and as our industry evolves these patterns are evolving as well. A major difference between software engineering and the analogous fields people use when talking about patterns is those industries are much older and move less quickly
If you are a language designer and you see lots of people writing the same boilerplate, it behooves you to put it into the language. A pattern is a desire path - pave it. In that sense, they are missing language features.
I believe C has allowed passing and returning functions from... the jump, no?
def addX(x: Int): Function[Int,Int] = {
y => x+y
}
Once you get used to doing this, you realize it's useful everywhere.
In a decent language with functional programming and generics support a lot of GoF patterns can be directly encoded as a simple type signature where you receive, return, or both some function, so there's not really much else to say about them. Like half of the behavioral patterns become variations of the interpreter pattern.
You can have your building engineered, in which case building walls out of 2x6's 16 inches on center is not off the table, but neither is a mortise and tenon timber frame with partition walls. In that paradigm, the code tries not to be descriptive of an exact technique but only gives you criteria to satisfy. For example you could run all of your electrical wiring on the outside of the walls or on the outside of the building, and you could use ramps instead of staircases. It only talks about ingress and egress for fire safety, and it explains how you're supposed to encase wires, or if wires are not encased it describes the way the wiring must be sheathed to protect the occupants.
You can heat your house entirely with an open fire, and the code speaks to how to do that safely. So it's unlike "design patterns" in a lot of ways in that the code tries to accommodate the kinds of buildings we try to build and the ways in which we modify buildings because that's easier than saying "these are all the allowed ways of building an entry staircase." Design Patterns are more in the latter category.
OOP has no firm theoretical foundation, unlike FP which is rooted in the formalisms of mathematics.
The first theoretical foundation of OOP is structural induction. If you design a class such that (1) the constructor enforces an invariant and (2) every public method maintains that invariant, then by induction it holds all the time. The access modifiers on methods help formalise and enforce that. You can do something similar in a functional language, or even in C if you're disciplined (especially with pointers), but it was an explicit design goal of the C++/Java/C# strand of OOP to anchor that in the language.
The second theoretical foundation is subtyping or Liskov substitution, a bit of simple category theory - which gets you things like contravariance on return types and various calculi depending on how your generics work. Unfortunately the C++ people decided to implement the idea with subclassing which turned out to be a mess, whereas interface subtyping gets you what you probably wanted in the first place, and still gives you formalisms like Array[T] <= Iterable[S] for any S >= T (or even X[T] <= Y[S] for S >= T and X[_] <= Y[_] if you define subtyping on functors). In Java nowadays you have a Consumer<T> that acts as a (side-effectful) function (T => void) but composes with a Consumer<? super T> to get the type system right [1].
Whether most Java/OOP programmers realise the second point is another question.
[1] https://docs.oracle.com/en/java/javase/21/docs/api/java.base...
Except dependency injection. I really can’t imagine why you’d ever not use that. I suppose it’s possible to overuse, but you’d still have better code than without. Certainly more testable code.
With direct dependencies, if you are trying to understand some code that calls some function and what it does exactly isn't completely obvious, you can press a button to go to it, understand it, and come back.
With dependency injection it depends on what is going to be inserted during runtime, so you can't.
Hence the benefit to testing; allowing you to inject a deterministic implementation while under test.
Dependency injection simply means to take objects as parameters, and not instantiate them themselves (which causes "Inversion of Control" also commonly mentioned when talking about DI). DI Containers just makes the managing of objects easier.
Avoiding it like a plague seems excessive, did you have a bad experience with them?
Design patents are more of "you need to build house with this exact room layout" than "the materials and ways to put them together are standarized"
I see you're familiar with Uncle Bob's handiwork
OOP can be wonderful, but the people who aren't able to step up a level in conceptual abstraction should really not touch it. Remember, for many years languages like Java didn't have any concept of lambda's and higher order functions, so design patterns were essential for elegant solutions. As they say, a design pattern is a symptom of the language being not expressive enough. In other words, many design patterns in OOP languages express the same thing as first-class language features in the functional paradigm would do, Visitor vs fold for instance.
They're the corporate equivalent of USSR soviet style conformism, when everyone had to call each other comrade and refusal to do that had repercussions.
Similarly, if you say you refuse to follow the Agile/Scrum manifesto or clean code practices, you get ousted, as that's Haram/not-Kosher in this racket.
I still wonder how Valve manage to ship Half Life without Agile or clean code practices.