Hacker News Reader

Comment by 01100011 | Hacker News Reader

I periodically work on c++ instead of C and each time it follows a similar pattern where I learn about some new c++ trick, think it will make things better, write my code, hit a compiler error, then spend the rest of the day learning why I can't do what I want. Granted, I usually am stuck at c++14, and some of the issues are fixed in future versions but still...

I really want to love C++. It gives me a more powerful C that theoretically should improve my output but in the end it carries so much cognitive baggage that I usually end up writing "C with classes" and move on.

View on HN

> then spend the rest of the day learning why I can't do what I want

There's a point when learning's fun, I think the OP is still there.

I wrote a bunch of realtime C++ in 2003, hated it. But last year, I wrote most of my code in C++ and liked it finally.

Lambda and auto was the tipping point where I stopped hating it. Some templates thrown in, but mostly to avoid writing a lot of branches for variants of the same function.

With lambdas I could now write the way I was initially taught programming, mostly by Lispy teachers channeling SICP.

Didn't hate the allocations either with unique_ptr.

"Lispy teachers channeling SICP" made new audibly laugh

If you haven’t already look up ‘if constexpr’ in cpp17 and newer. It lets you have compile time branches in a single function.

Also requires, which initially looks batshit insane but is in fact really quite cool: https://www.think-cell.com/en/career/devblog/if-constexpr-re...

> Unlike requires requires and requires { requires }, which are perfectly reasonable C++ code, requires requires { requires } is completely silly.

Normal day in cpp land...

How on earth did C++ brainrot lead them to think that reusing that keyword was anything but a nightmare? The very first line of the article

> Probably the two most useful features added to C++20 are requires and requires

said without a trace of irony ... the whole thing reads like parody!

I keep hearing that C++20 or whatever is getting so great but if so why would they specify such a dumpster fire of syntax???

> There's a point when learning's fun, I think the OP is still there.

What a horrible idea that one might no longer be there.

You hiring for positions where I can have fun programming?

Never hired anyone. But also never worked in any place where programming was not fun for me. But also never touched any Java or C++ job.

I see a lot of bad code from people who have just learned about some feature and want to use it. Don't do that. Think of how to write your code, and then, if a feature solves a problem, use it. Problem->Solution, not the other way around.

E.g. C++ templates are generally pretty awful, but sometimes, compile-time duck typing, or automatically adding padding based on sizeof(), etc, is very useful.

This describes my C++ experience precisely. Initially I was a C# programmer learning C for embedded projects and it went exactly like this.

C++ has a lot of really neat features that sure look powerful if your application aligns precisely. It seems like every time I try a new feature, what I want to do is always an edge case that doesn't work in my situation. I try for a few hours/days before I give up and just write it in C.

My biggest disappointment with C++ is that the standard libraries are completely unsuitable for use in embedded systems where you have to control when, where and how many memory allocations occur. This is particularly important when the system level design choice is to perform all allocations at system startup, which is a common design pattern for high performance systems with real time characteristics. A high speed messaging system I worked on ran into this all the time. We couldn't use the standard implementations of things like heaps, hashes or queues because they don't have a way of making memory allocations occur at startup. It was quite common to have to re-implement those data structures when adding a new feature, as the stl or boost implementations were not suitable for this design pattern.

Have you tried override the allocators and destructors of your classes so you can do all allocations at startup?

C++, like most technology, works better if you put in the work to learn about it.

(Admittedly there are languages like python and ruby that buck this trend.)

> C++, like most technology, works better if you put in the work to learn about it.

There's also the impact of software entropy: if someone has little to no experience developing software and has to grow it by adding features and fixing bugs, over time their contribution to the project invariably results in degrading it beyond the point they can possibly salvage it.

At that point, they blame the tools.

Sure, but the effort/reward ratio bears some consideration. I've put a lot more effort into learning C++ than any other programming language but I'd still say it's just the language I'm maybe 4th most proficient in.

Programming languages are a means to an end. I'd rather have my cognition going towards solving the actual problem and not worrying about implementation details.

Python does work better if you put in that kind of work. Otherwise you'll get bit by the way mutable default arguments work[0] (or never learn to use it to your advantage), or by late binding of closures[1] (and maybe you'll pick up the awful habit of exploiting the confusing early binding of default arguments to make the `lambda`s that you constructed in a loop work properly[2]). Or you won't get the big picture of the descriptor protocol[3] and how method lookup[4] is related to the implementation of properties[5]. Or you won't get the fancy metaclass[6] thing that your favourite framework is expecting you to treat as an opaque abstraction[7], or how and when you might use a class decorator[8] instead of a metaclass to do that kind of metaprogramming.

[0]: https://stackoverflow.com/questions/1132941

[1]: https://stackoverflow.com/questions/2295290

[2]: https://stackoverflow.com/questions/3431676

[3]: https://docs.python.org/3/howto/descriptor.html

[4]: https://eev.ee/blog/2013/03/03/the-controller-pattern-is-awf... (yes, this was partly an excuse to get one of Zed Shaw's critics [9] into the discussion)

[5]: https://stackoverflow.com/questions/3798835

[6]: https://stackoverflow.com/questions/100003

[7]: e.g. https://medium.com/@miguel.amezola/demystifying-python-metac... - I didn't have a good link for this and didn't spend a long time searching, but this seems okay

[8]: https://stackoverflow.com/questions/681953

[9]: https://eev.ee/blog/2016/11/23/a-rebuttal-for-python-3/

You are forgetting these teeny tiny teensy little things called "return of investment" and the ratio of energy expended vs. the results achieved.

Python has a lot of footguns (and special __ methods), and insane, alien, scoping rules. And lots and lots of syntax, as well.

It is not easy to learn (I have no idea why it's considered a beginner friendly language). It benefits from putting in work to learn it. It's not as bad as c++, but it is not a shining example of how to get it right either.

chatgpt or copilot helps a lot with that. I usually copy the text and let it decipher the thing for me.

they’re still bad for generating any significant body of code.

They’re a little bit better about deciphering errors.

They’ll still bullshit* you and send you on wild goose chases.

*hallucinate if you prefer

> They’ll still bullshit you and send you on wild goose chases

And confidently at that. It can't seem to find the backbone to say no to me either.

If I say something like "wait, X doesn't seem to make sense, isn't it actually Y and Z?" it will agree and reformulate the answer as if Y and Z were correct just to placate me. I usually use the LLM to learn new things, I don't don't actually know if Y and Z apply.

Reply on news.ycombinator.com