Details
-
AboutBackend developer / system admin
-
SkillsC++, C#, Java, PHP, Kotlin, Python, HTML, CSS, JS, Unix, Apache, some docker and lxc
-
Website
-
Github
Joined devRant on 3/18/2019
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
-
@Wisecrack Man, I hope it's gonna work. One of the worst parts about this whole LLM hype is the amount of resources it requires. The second worst part is that big tech firms are using it to monopolize the market with subpar products that require even more resources and are literally making life miserable for people while also objectively hurting the environment without a second thought.
I really think we fucked up big way with LLMs. The capabilities it have are nowhere near what they need to be to justify the cost. This is literally a kids toy that tech giants can't stop playing with and everyone is paying the price. I'm confident these models can be way more memory and power efficient and we need to democratize this before we're all running GPU farms at home... -
@Wisecrack When I was stuck as a young dev, I would often find a solution on SO and deliberately copy it by hand because I wanted to *learn* the solution better. Surprisingly enough, this actually differs from just copy-pasting it because as you write it, you think more about it. It happens often that when I start rewriting it code that looked good at first sight I will actually discover things about it that could be done better in my case, or things that are really clever tricks I didn't know about before that didn't jump out immediately upon just reading, but they did during careful copying.
I think this is another skill that LLMs will take away from devs, making them objectively worse as a result if they are not careful :D -
@retoor LLMs are *not* capable of creativity. If you have steered it into a new and interesting direction that's *your* creativity. Context and probability is everything to an LLM. The rest is Apophenia.
I personally don't hate LLMs, but I strongly dislike the cult of lies that's been built around them. We don't need a definition of slop as urgently as we need peoples technical understanding to increase regarding these tools. You can't solve this with more doublespeak and buzzwords. -
@retoor your definition of "ai slop" is close to mine, but I personally think the effort put into the prompt doesn't mean the result can't be slop. I think output quality control should be a major factor. Whether something is slop or not should be a function of both the output quality itself + the effort put into increasing the quality.
Someone can have an AI farm where they spent months optimizing the prompt, and still the AI farm is then used to produce millions of similar, bad outputs. Essentially a long and elaborate spam email isn't any better than a short low-effort spam email. It has to be something more to not qualify as spam anymore.
I'm sure there's a deeper philosophical debate here and one could probably write a 100 page thesis on this topic... -
Slop is slop. Bad code is bad code, bad art is bad art, effortless crap is effortless crap.
AI slop is everything that's slop, but made by AI instead of a human.
There's another word used for slop made by humans specifically in the software world: "Shovelware".
But in general you can think of AI slop as being low effort, low quality, high quantity, without thought, unfinished, half-assed...
The issue is that almost all AI generated content is slop by default. It only looks right on surface level (especially in visual art you can see the horrible artifacting, nonsense shapes and smeared oil-coated detailing). That's slop. In code it's harder to spot, but you can still see it in larger projects, especially if the person didn't know what they were doing or if there was a lack of direction and scrutiny.
As a rule of thumb, if it took 15 seconds to make, and it also looks like it, it's slop, AI or otherwise. -
504 lines seems a bit misleading, since things are just formatted to be one-liners and pretty much all the variables are really badly named. ^^; This reads more like obfuscated code than readable code.
In fact formatting this according to the pep8 rules yields 727 lines. Though sometimes the single-line formatting does work nicely and I don't hate it.
I think it could be simplified further though. This function irks me:
```
def is_truthy(s, v):
if v is None:
return False
if isinstance(v, (int, float)):
return v != 0
if isinstance(v, (str, list)):
return len(v) > 0
return True
```
especially since this is python you don't have to care about types *that much* I think it could be implemented like this:
```
def is_truthy(s, v):
return v not in ["", 0, [], {}, None]
```
the typeof can also be simplifed, but I'm out of chars -
nothing merged into master should break master. That's the only real requirement. All other requirements are fluid and change based on if the deadline is in literally 10 minutes or in 5 months. But the code has to compile and must work once in master.
Everything you were forced to ignore in that request becomes Tech Debt and will be fixed in the next release, or more likely never ;) -
I know this one! The answer is Windows XP on a 2001 machine with an HDD.
I timed mine back then and it took anywhere from 8-13 minutes to boot completely.
Man, glad those days are over. Booting from an SSD is less than 30 seconds and it would take a special kind or idiot to fuck that up on an OS level -
@whimsical I like to think of myself as somewhat of an educator. Give a man a correct answer, and you helped him with one problem, but teach a man how to get their own answers and you have to answer less questions, or something... :D
Ofc in practice rarely anyone cares, but oh well. -
Game theory in action
If everyone had the "fuck you" mentality. Companies would be forced to update their hiring proceses, otherwise they would have no workers.
But since you can't be sure that everyone will adopt that mentality (e.g. a snitch). You are risking by adopting it, even though you'd all be objectively better by not working at these companies, you can't risk not having a job due to snitches.
Game Theory tells you what is objectively the best option for everyone, but it also tells you why It's the worst choice for you.
Long story short, you should adapt the fuck you mentality only if you can afford not having a job (e.g. if you already have some sort or income)
Otherwise the best option is to play the game as is, unfortunately. -
Is this why my team reported github being down yesterday?!
-
wat, you can create tables using API calls? why? what? how is that secured? can anyone? what? what about migrations in the future? I never heard anything this dumb before... Is he a vibe coder?
-
"vibe coders" tend to forget that sofrware engineering is an "engineering" job. The code itself isnt as important as the decisions made, the tools used, the flow executed, systems inpacted, users helped... There's a whole lot to our job than just writing code that works. Vibe coding is the antithesis to good design. Code that works isnt always the best code for It's purpose, deployment site, user interface etc...
That being said. If you find exactly the perfect code on stack overflow, or if you bash it out of a text generator, or hell, if you find it scribbled on a toilet stall and rewrite it character by character to your codebase, It's not going to change anything, because as I just said, the code itself is secondary to It's purpose and design. No one can stop you from copying a solution to save time if you would've writen the same one. Works smart, not hard always applies. Just make sure you know exactly what you're making -
The only important thing is that you understand every line of code and the decision behind it.
The problem is blind usage of SO and AI, which you just copy. Even if you understand what it does. That doesn't mean it makes sense to code something that way, or that it fits the rest of the project, or that it's even functional code.
We had this requirement too. This was before AI was a thing, so this applied mostly to copying code or cheating. The way my prof. checked is that he interviewed everyone on their project and you had to be able not only to understand what the code does, but also why/why not different code or placement. He knew if you wrote it yourself because if you did the actual research you would likely land on something good, and even if your code was subpar you'd have strong opinions on why you did it that way.
He was a great teacher to me and he took it very seriously. He was one of those guys that would rather hear a bad answer than no answer at all. -
What part of that is surprising considering the history of big corporations? Literally the one strategy they use over and over again
-
"He"?
-
Micro Services can be done right, but not with lambdas. And the only way microservices end up working well if they are separated by strict domains and there's not a lot of them. In any project you're unlikely to need more than 2-3 microservices. If you go beyond that there's a good chance your domains are split too small and the communication overhead will kill all the benefits.
And lambdas are by design too small to do this. They are just functions in a cloud and that's just hell. It's literal hell. Imposible to debug, track or test. Especially once you start handling communication via queues, dynamo updates and event triggers.
It sounds nice on paper, and it might work for very basic projects, but maaan, it doesn't scale at all. I mean it scales performance-wise but not maintenance wise -
@12bitfloat hey, maybe it wasn't get "dumb" decision. Maybe this guy is a control freak that breaks into her phone. Who knows what other privacy basics he disregarded! Girl went to a friend, spilled her heart "yeah, OP just keeps breathing down my neck, I don't think I can stay in this relationship any longer" the hooked up and now we're here!
OP came in with a name like "sweetkid" and a fresh new account. So it's either a troll like Sid or someone, or it's a guy that legit thinks of himself as "the nice guy" and "the victim" and made the relationship toxic himself... Eitherway, fuck a guy who's first reaction is "how to get into my gfs phone". -
not sure if I misunderstood, but I already do processing like this in some of apps on a higher level. Isn't this pretty much how SIMD architectures handle stuff anyway? You load a bunch of data into the memory, and then each thread operators on it's own bit/stride of the data.
Not sure if this is different, or if you want to make this into a more low-level concept in compilers or in hardware or something, but as far as I can tell the technique isn't new as a concept... unless I'm not getting it :D -
Besides, I don't care for these companies one bit. I'd rather see this tech firmly in the hands of the open source community than a fake non-profit.
-
Yeah, while we live in a world where people have no critical thinking and end up hooking up in relationships with AI, and killing themselves or others because of AI, or literally spiral into full on psychosis because of AI, I'm fine with knee-capping AI. In fact I think those corporations should be held liable for what their tools and misinformation cause.
the tool is one thing, but the way they chose to sell it, that's on them. And they managed to convince a big chunk of people that this is sentience and essentially magic, so they are on the hook.
If you sell a radioactive rock to someone, saying that it's a magic rock that will fulfill your wishes, and hundreds of people die because of it, then yeah, the people were probably on the way out of the genepool already, but you still did that wanton disregard...
When the AI craze started and we only had bits and pieces of information, it made sense, but at this point it's criminal negligence in my opinion -
@whimsical no, It's OpenAIs fault for lying about the capabilities and convincing people it can do what it cannot. You're right
-
@Lensflare It does, though it usually tries to hide the specifics of it from you. It's mostly trying to do the C# thing where you either pass by value (default) or pass by reference (explicit in C# you'd use the "ref" keyword, in Go you just use the "*" symbol, so feels similar to pointers in C)
however you can't do pointer arithmetic on those references unless you go out of your way to specifically cast it to an unsafe pointer type. Infact there's an entire package called "unsafe" dedicated to very low level access.
Honestly, I've been working on and off with Go for about 2 years now? And I like many things it does. There's still a bunch of things I'd like done differently, and it's a very verbose language due to it's error passing style, but for many things it's just very efficient and easy to work with :D -
Golang does that. When you're working with a pointer it dereferences via dot and if It's a value copy (which is the default behavior) it accesses the value member.
So It's definitely doable to use the dot for everything. I think the reason why C's approach might be better is that it's explicit.
Explicit is always better (as in, less error prone) than implicit, so I get it.
It's not a big issue in Go eitherway. -
If you start all your arguments with "Listen bitch..." then you are guaranteed to get more attention.
Otherwise, dunno. You just need to be really sure what you're talking about, exert confidence and demand to get answers or actions on your concerns. It's a plus if you ask specific people by name. You should also make sure you get a specific slot to speak up and then say what you need to while people are listening.
I'm not exactly sure why people listen to me, but more often than not I don't really give them an option to ignore my requests. Especially naming specific people or asking "what do you guys think" can prompt answers or sub-questions like "sorry, I wasn't listening, what are we talking about?" :D -
Maybe it was rsyncs fault. It does a lot of things and It's kinda optimized for remote copy.
Did you make sure encryption and compression are disabled? Used whole files?
Maybe dd would be better if you want a whole disk clone rather than just some folders -
Just to add to previous comments. There's also usually some self-discovery going on. You can be unsure of your orientation and experiment. You might think you are gay for a while, until you meet an experience that negates that. Emotions are complex and sometimes you can feel confusing things towards people, especially if there is some sort of stress or trauma involved.
Being "gay" may be a binary label, but the self discovery is a Journey.
(Which is why It was so annoying when everyone glorified homosexuality for a while, and still do with that whole pride month bullshit. it made an already confusing journey even more confusing and entangled with politics and peer pressure) -
Interesting, elaborate.
-
I just noticed a bug in a PR recently too. When I brought it up to the mate who wrote it I got the response "copilot wrote that".
:)))))))
I like my colleagues and all that, skilled people and good culture, but damn, stuff like this raises my blood pressure -
Half my home infra is relient on vpn and iptables. I agree that it can be difficult, I essentially have to re-learn it every time Im doing something with it. But It's not that complicated. It simply gives you hooks to every part or routing inside the system and filters to work with it.
There's probably an UI for it somewhere on GitHub though. What I do is usually make a new shell script with "up" and "down" options so that I can work with it and revert it easily during develoment and testing
