26
I think LLM coding tools kill the need for 99% of language fluency and expertise (hexbear.net)
submitted 1 week ago* (last edited 1 week ago) by [email protected] to c/[email protected]
43 comments fedilink hide all child comments

Just ask it to rewrite the shitty code you wrote in a language you barely understand to "follow standard best practices in " or something like that and it will add advanced typing features, functional programming for iterables, advanced exception handling, proper concurrency handling, optimize control flows, use better equivalent functions, etc.

As long as you understand the foundations of these concepts in at least one language anybody can become pretty close to an expert in most languages instantly. Especially since most of them are C based and pretty similar

The output will sometimes change the logic but I mean that's pretty easy to catch and fix

Rip C++ nerds that memorize the entirety of each releases manual to shave off 3ms in every single function

you are viewing a single comment's thread
view the rest of the comments
[-] [email protected] 20 points 1 week ago* (last edited 1 week ago)

Become “pretty close to an expert” by… outsourcing the process of improving your code to a machine…

Even if it improves your code in that scenario, you’re not going to really understand what it’s doing or why. You can use AI as a shortcut for scripting, but you can’t use it as a shortcut for learning

Edit: Besides, we already have perfectly good static analysis tools. Just use a linter. Trying to use AI as a linter will just be worse and unpredictable compared to using an actual linter

[-] [email protected] 1 points 1 week ago

But I'm not using it for learning. I already understand exception handling, concurrency, typing, etc.

But I only know the exact syntax for some languages

Now I can replicate the best practices for those concepts in a language I've never touched, and I can understand what it does because I know the equivalent syntax in another language and so I can also judge the quality as well

It's even more confident when the new language is C based because I'm already familiar with other C based languages

Obviously it'll never be as good as a person who spent time to learn the language by reading documentation and practicing but most cases it should be fine

[-] [email protected] 18 points 1 week ago

I already understand exception handling, concurrency, typing, etc.

But I only know the exact syntax for some languages

Now I can replicate the best practices for those concepts in a language I’ve never touched, and I can understand what it does because I know the equivalent syntax in another language and so I can also judge the quality as well

I'm sorry but this just doesn't match my experience. I have used greenlet, Node.js, asyncio, POSIX threads, kqueue, and uv and just recently I had to look at something that uses tokio (Rust) and I would never say confidently that just because I know the syntax of one concurrency library, that looking at a different language and equivalent library I can immediately judge the quality and understand what it does.

That is just not realistic

[-] [email protected] 11 points 1 week ago

Don't you know that all programming languages are the same, except each uses a different set of symbols and keywords? Since all languages are the same, we can use an LLM to efficiently translate code from one language into another where it will perform optimally. /s

[-] [email protected] 7 points 1 week ago

"Computer, replace all the whitespace indentation with curly braces and put a semicolon at the end of every line, in order to convert my Python program to Rust"

[-] [email protected] 1 points 6 days ago

Python also allows you to override what operators do meaning I can write

int.__add__ = lambda self, other: print("hello world")

>>> 1 + 2
'hello world'

And that's totally valid code. Don't think many other languages allow that, and translating that would be a mess.

[-] [email protected] 2 points 5 days ago* (last edited 5 days ago)

It's technically operator overloading, the wacky thing that Python allows you to do, is overload operators for base types like int which I'm not sure if other languages allow you to do for base types.

[-] [email protected] 2 points 5 days ago

That's what I meant, you can modify the behavior of code by directly over riding the operator implementation for base types. What it really reveals is that Python int is not at all a C int or really any other int.

Directly translating syntax without knowing that the Python type is so vastly different from say, the C type is a recipe for latent disaster.

[-] [email protected] 2 points 5 days ago

This is true.

[-] [email protected] 1 points 5 days ago

My favorite weird Cpython implementation detail is that -5 to 256 are pre-cached when the interpreter is initialized. So identity checks using those numbers return True, but return False for other numbers:

>>> x = 1
>>> y = 1
>>> x is y
True

>>> x = 100000
>>> y = 100000
>>> x is y
False

At least in newer versions of Python it screams at you for doing identity checks with integers

[-] [email protected] 2 points 5 days ago* (last edited 5 days ago)

Yes, but that is a common optimization, caching primitive types and common values. I believe the JVM has that behavior, as well as a couple Ruby implementations (MRI, possibly YARV but it's been a while since I looked at Ruby implementations)

[-] [email protected] 2 points 4 days ago

Yeah, cashing the 8 bit values means using integer flags is a lot faster. Implementation details like that are always a kicker. Especially when Python's syntax kinda makes you want to say "val is 1" since it's more "human readable" than "val == 1".

This is actually abused for True and False too which are a subclass of integer with Sentinel values equal to 0 and 1.

Since they're technically different objects, True is 1 is False, but int(True) is 1 is True. True == 1 is True though.

Basically every language with an interpreter will do stuff like this, but it's usually not super well documented behavior as it's considered implementation detail or private API stuff.

[-] [email protected] 2 points 4 days ago* (last edited 4 days ago)

Python’s syntax kinda makes you want to say “val is 1”

I have absolutely had issues with this. My editor configuration python-mode raises an error about this via flake8 but ruff ~~doesn't~~ didn't raise an error about this (and I have enabled experimental settings) so my co-worker added code like this and it slipped through.

Ruff now has a rule for it but I think I was a very early adopter where this hadn't been implemented yet.

https://docs.astral.sh/ruff/rules/is-literal/

[-] [email protected] 1 points 4 days ago* (last edited 4 days ago)

Python ~~3.13~~ 3.8+ now spits out a warning to stdout when you do this, it was a big enough issue that they had to bake a warning into the CPython interpreter lol.

But because it's a big enough issue, it also means there's probably tons of examples online of people writing code this way and that'll be something you have to be actively watching for when using generated code.

The reasoning models would probably ping on some forum post about it, but because it's technically valid Python code, there's a good chance that it would see enough identity comparisons to integer literals to think it's okay.

The hard part of programming always ends up being the undocumented implementation details like that. Syntax is easy once you can understand the structure, but it so easy to write something that technically works and passes the tests that will eventually fail spectacularly and be hard to find.

[-] [email protected] 2 points 4 days ago

100% agree with everything you've said. I think it really illustrates how Python has not really done a good job around memory management.

By which I mean, it has done it well enough where you don't have to think about memory management normally and that actually ends up hurting you because the identity operator explicitly is about checking if two references point to the same object and people forget that, and then some interpreter optimizations around small integers make it so the is operator and equality operator behave the same for small values and people get a wrong impression about the identity operator.

I have the same problem with Python's "pass by assignment" way of doing things where it's not explicit enough for my liking. Don't get me wrong, I understand what it is doing after 20 years using the language but I sort of appreciate C like languages where references vs values are more explicit with things like ponters.

Zig has been interesting to me, for this reason.

[-] [email protected] 2 points 4 days ago

Haven't gotten into Zig, but it's on my list. I'm trying to get into Go and Rust now. Especially since I too have almost 20 years of using Python under my belt and hit those annoying "sometimes I just want to be explicit" moments.

That being said I will always come back to Python especially because I do now understand those pitfalls that can totally stump someone who's new to the language. The general syntax and elegance of generator expressions is addictive for small projects and quick tools. No boilerplate and a few comprehensions can do the work of a whole library (albeit quite a bit slower, but pretty frequently runtime is secondary to maintenance time).

I do like that type hinting is becoming the standard in Python though. I have absolutely abused the private apis for reading type hints in my code to get a sorta poor man's runtime type validation in mission critical (database) fuctions. I've used type hints with a decorator to build API endpoints as well. Hope that someday Python finally just commits and stops pretending to be lazy Rust and just allows you to opt into static typing.

[-] [email protected] 7 points 1 week ago* (last edited 1 week ago)

In case my edit didn’t land in time: what makes the AI approach better than using existing non-AI static analysis tools

[-] [email protected] 2 points 1 week ago* (last edited 1 week ago)

Well my personal experiences have just been that the ML approach catches a lot of things the static analysis tool hasn't. Those are hard coded by humans and there are dozens of not hundreds of ways to write any given function with identical logic. It's impossible for static analysis to be comprehensive enough to catch and fix a code block more than a few lines

E.g. I had a super ugly nested try catch exception block for this new integration test I was writing. It was using a test framework and language I've never used before, and so it was the only way I knew to write this logic. I asked the LLM to improve it and it broke up the nested try catch logic into 2 top level pieces with timeout functions and assertion checks I didn't know existed. The timeout removing the need to throw an exception and the assertion fixing the issue of catching it

[-] [email protected] 9 points 1 week ago

I’m glad you’ve gotten some actual use out of the LLMs! My outlook is more skeptical because I’ve seen too many interns get stuck on projects because they tried to ask LLMs for advice (which they did not double check) instead of reaching out to an experienced dev. The word calculators can only do so much.

[-] [email protected] 2 points 1 week ago

Oh don't get me wrong, I definitely think LLMs are gonna absolutely destroy kids ability to learn anything, including coding if they use it like a teacher

But for those who use it as a tool to build and do instead of learning, I'm quickly starting to become a strong believer in its usefulness

this post was submitted on 29 May 2025
26 points (81.0% liked)

technology

23805 readers
318 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 4 years ago
MODERATORS
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]