True ! There are many blades in the swiss army knife, if it's a screw, use the screwdriver blade and stop blaming the knife. From a teacher's perspective it's hard: students are excited to use/show their knowledge and it's normal. Normally, soon they become wiser, but not always.

Many programmers (such as myself some years ago) mistake simple code to "layered code". A truly simple code is not one where you have multiple layers of abstraction hiding complex inner logic, but one where the logic is totally explicit and straightforward.

Nothing personal but that's a dumb take. Not everyone is a fucking quant dev and needs top speed. On the same article, the guy shows another example where ranges let you write cleaner code. The point is to benchmark your code and to not marry a tool. Use what you need.

This is actually very beautiful, anyone who coded in a functional language before will appreciate it.

Having this abstraction available for whenever it is needed isn’t bad at all. The language doesn’t force you to use any abstraction over the other. It’s your use case that does. And to have 2 different abstractions in the toolbox is not had thing. The bad things in this snippet are: the example is horrible, and the C++ legacy syntax makes such abstraction feel a little out of place.

accurate

it's ironic. C was powerful for keeping the language as simple as possible == moving as much as possible out of the language, into headers and libraries

The ranges library, ( opt in btw ) sacrifices a little performance for functional programming. It shines when composing algorithms from the stl library.

It is a rat race in who will be writing the most unreadable APIs!

Most people just follow philosophy of the language and they will use language features just because they spend time to learn them. "I paid for the whole speedometer, I will use the whole speedometer".

I was using C++ 14 but now converted all to C99 it didnt took much time. Code is simplified I removed much more than I've added.

It’s just hold over from pre-SIMD days. They’re still coding like it’s 1999 avoiding loops at all costs lol

I think the complexity and flexibility of either approaches are subjective and providing abstractions alongside the raw equivalents is ideal.

Well said. I think the psychology of it is that we like cool things, exciting, glamorous, and we often forget the simplicity is quite often not so "cool".

That's just functional idiom support. No inherent additional complexity

Hello imgui guy

iirc main idea behind cpp was that its c with oop. and I fucjing hate std library

That is a great example of trying to be something that you're not. Trying to be Rust but it is not.

The abstractions should serve a purpose. You do it for some function. Haskell people have list comprehensions because you can compose them from disparate sections of the codebase. The C++ code is entirely form with the same function as a loop, is it not?

I find the use of auto here the most offending, actually

you're talking about eric?

This one might not be a good example of this marvelous technology, but I once was so tired implementing algorithms involving tree traversals so I created a (pre-order) tree traversal iterator. And ranges are a nice concept for a composable API!

Yet this abstraction is less complex and more flexible. The idea of a filter seems pretty upfront. Yet another alternative would be std(::ranges)::copy_if, which specifies what you’re doing (good) but requires explicit declaration of the output vector.

Modern C++ is just a joke. And a pain in the ass to debug code that is not yours. Coroutines should never had arrived to C++.