#python-discussion
1 messages · Page 452 of 1
they prbly learnt at least one thing
Yeah its hard to understand but im reading it
metaclases are basically the type of classes
classes of classes as type and class are interchangeable
the built-in default metaclass is type
really the most important part is
i think they are a topic which is just interesting to know about to understand the object model and hierarchy of python better, but finding a real problem which they solve is very rare
at least in python
you can absolutely never learn about metaclasses and still use python effectively
they are pretty much an esoteric feature
(in most cases b4 someone finds the one case)
!mute 1242525097357738095 1D Seriously take a break. You're not actually reading my messages or understanding what I'm trying to say.
Misleading beginners until someone else has to step in is not okay, even if you mean it as a joke. I've told you before that tone is hard to tell via text. If you do not make it clear, then it's not creating the environment we want. You have to follow our code of conduct in this server.
Take a break and think about how you're coming across.
:incoming_envelope: :ok_hand: applied timeout to @fossil steeple until <t:1775857942:f> (1 day).
guys im sorry so i hade state testing and i dind't study nothing
in python and i forgot stuff
i learned before
what i meant by 1 : int : type is that just as how 1 is an int, int is a type
and just like how you can define something like int, you can define something like type, and instances of that would be something like int
Ohh.Hard to understand 😅 .But thank you so much bro,I appreciate it
its absolutely not necessary to understand, specially for a beginner
but if you like learning about unusual language features, its one of those
if you do want to learn about it - you should only learn about it later, as it is impossible to actually grasp it before learning atleast basics of "OOP" in python.
What did you forget?
My next lesson is.
Variables
Data entry
Follow the values
Program the calculations!
i think topics like metaclasses are what separates "the working programmer" from a "programming language enthusiast"
they are like, absolutely not necessary, and almost impossible to find a usecase for, but just interesting to learn about
I'm closing in on a decade of python experience and have never used them.
decadeish
though knowing that in python types are first class values is pretty useful i think
like, you can just have code that works on types to do some basic but useful metaprogramming
The keyword def is used to define functions. Functions are blocks of code that you give a name. You can define parameters for the function which are variables that will exist within that function's scope. When you call a function you specify the function name and any arguments that the function will take in parenthesis. These arguments must match the parameter list that you defined for the function. You can also return a value from a function. The return value must be assigned to a variable to prevent it from being lost. It is also possible for a function to return nothing, and in that case the return keyword isn't necessary. A simple example of a function that takes one argument to fill one defined parameter and returns an int value is:
def foo(x):
return x + 3
Which you would call like this:
result = foo(5)
And you can print the returned value you assigned to result like this:
print(result)
This would print the value 8 to standard output.
I have never done anything with metaclasses outside of a classroom.
ideal; you've probably been making things that are actually useful /j
Wait. What does the /j mean in this context 😄
They can be useful but not too often.
Just kidding. Or joking.
it looks like I'm joking about them making useful things
I don't mean that 🥺

I interpreted it as you saying making metaclasses is less than useful.
that was my intention
In a joking manner.
I am just teasing, lol
!warn 1392976550722273290 Don't advertise here
:incoming_envelope: :ok_hand: applied warning to @runic hound.
@grave treehow
omg I thought soph got warned at first, since I onyl saw it out of the corner of my eye
This server is not the place to advertise your youtube channel
You asked people to subscribe to a youtube channel. That's a form of advertising.
yes, that's the same as advertising. You're trying to get people to subscribe to your youtube channel and posted a link to it. That's advertising
paradoxically I find tone tags often obscure meaning more than they reveal it
I start second guessing why they even decided to add it
Text is not an ideal form of transmitting nuance.
[Resignated statement] We really need to just start talking like Elcor
Elcor???
you're right, we should use stage directions instead
They were a race in Mass Effect series. A lot of their communication was based in micro changes in their posture/gut fauna/smells or something that were indecipherable to most. So they always prefaced their statement with something like "Chastinging rebuke. That's unfair to the new person"
Ahh.
Are people still buying stuff with their own money in the big 26?
genuine question
gas, probably
what's the alternative?
Are you offering to buy things for me with your money?
ever hear of BNPL?
It's still your money in the end.
everyone's on that stuff now
That is the kind of thing that bites people in the ass.
couldn't say less, bit my aunt in hers.
Anyone here planning on going to PyCon US?
i cant even afford a ticket to the US
a cricket to the US
Apparently.
The proliferation of BNPL is really concerning.
Do I want to know what it is?
I mean, I guess I want to know enough to make that question
Buy now pay later.
klarna
i'm not convinced bnpl is done even remotely as often as it's offered
Black hat python api end point hacking with security vulnerabilities💔
well yeah
theres something called rent and loans
Steal
Its done enough to cause 100s of millions in net losses for klarna
(People don't pay it back)
I guess you could use free software if you don't want to pay
hire leonel and marco
Black hat
As a reminder for folks, this channel is for Python discussion and also our #rules do need to be followed
💔🙏
I ruined my black hat by washing it with bleach. Now I'm a whitehat.
Ask me how I turned it into a redhat
Systemd
my white hat has an oil spot that didn't go away 😔
so how goes pdm
Use the "everybody loves systemd" shirt
what's pdm?
I switched most of my repos to use Justfile
kat rebelled against uv about 20ish hours ago
it goes well. The init is too verbose for my tastes, but hey, survey result stuff is underway
Is that gnu make but MIT?
Its based on make
But it's not a build orchestrator
Ninja is what you should be using
yeah i'm hoping uv cleans up it's act before my next project... but dang openai didn't waste any time polluting the waters
what are the merits of the rebellion?
What happend?
Where's that uv fork that implemented dev scripts?
openai bought astral, I don't think there's any damning evidence that they broke uv, but uv init is failing now
wait you too? 
i dont think theres a correlation
I haven't tried, was just responding to mud
i am petty, but their basic command to set things up in a existing directory failed with an error despite me following their guide. That annoyed me enough that i just won't use it now
Anyone who makes an LLC for their project probably plans on selling it.
Or they're trying to cover an employment gap on their resume 💀
the error message was clear, but i dislike the assumption that command made and that it was out of sync with their literal get started huide
have you seen the uptime chart of github yet
githubs uptime has always been ass
If I follow a guide someone made for their product and the guide doesn't work my confidence would plummet, too
Fork uv and put it on codeberg
real
What's github's uptime? 6
five nines? more like five zeros amirite microslop 😆
Not 6 9s, 6.
my cat nando says hello
there is no way OpenAI has affected uv yet.
- straightish line
- copilot added
- line resembles a seismograph
the cat actually wrote the github uptime %
I get ratelimited for going to the second page of search results
Engindeer
aside from the normal disruption a merger causes perhaps, but I didn't realize that kat's protest was as isolated as it is.
i've gotten 429'd on the first page before 😔
💀
the acquisition hasn't happened yet.
is there a public date yet
i haven't heard one
LOL I was wondering what that was
I thought you just had STRONG opinions on github's uptime
Is anyone still using github for anything other than work?
I yanked all my personal stuff and moved it over to codeberg with a local forgejo backup.
it's still by far the most popular git host for open source
Yeah, big projects but IDK I'd consider that work.
I was thinking more for their own personal bits and bobs.
i think the answer is still, it's the most popular.
Shame
is it?
It is to me, but I've got Strong Opinions (tm) about copilot and their opting people in to training their paid models.
if your repos are public, they're being used by all the AI labs.
True.
But not using private repos is basically something everyone has to take entirely on trust.
yes, github in theory has access to your private repos for training.
and selling to other slop companies
Supposedly they follow their https://github.com/customer-terms/github-data-protection-agreement
Sure, but what's the actual penalty if they get caught 'accidentally' scraping private repos? Essentially nothing.
so technically leads to copilot
i think we talked about it in here some time ago...
how would they get caught
I wished everyone moved to codeberg
why do you want that?
and whats the penalty is another thing as well, every corner of the internet gets scrapped constantly
what is done about it
verbatim reproduction of copyrighted materials, which has produced no significant consequences for any llm provider
offhand, no idea. But it's not without precedent. I think EA did that to Bethesda back in the 80s with one of the first Madden games?
blame the llm and adjust it to "fix" the issue then
the material gets added to a detection filter, yes
problem solved lol
(no actual clue about that incident btw...)
The idea that they have A Policy (tm) or a contract or whatever satisfies lawyers but I mean if the data is all right there and the probability of getting caught is really low and the consequences of getting caught is basically a budgetary rounding error... why wouldn't they?
yea exactly
But oops, already in our training set and we can't get it out so 🤷♂️
🤷♂️ won't train on such stuff no more pinky promise
Swearsies!
Can someone tldr/high level overview on the github situation? And is there a decent alternative?
if you want it private dont upload it i guess
do RNNs continuously take in inputs as the previous input is still being propagated through the network?
(sorry for off topic but i saw that there are no python discussions atm)
they have been unreliable; they are owned by Microsoft which is pushing copilot everywhere. Decent alternatives: codeberg, gitlab
other alternatives until bought by microsoft
🤷♂️
I might be out of date a bit, but one upshot to Gitlab was their self-hosted runner was just a docker container. Last time I checked, setting up a github self-hosted runner was a lot more involved?
Let's hope that antitrust does its job.
(keyword: hope)
No. It has sequential dependency. At least, if you're asking what I think you're asking.
The hidden state needs to be updated with every step in the sequence which creates a new hidden state.
There can be like.. batched/optimized ways of processing if we're talking about lower level implementation, but logically it still has to be sequential.
You might be thinking of transformers that can process time steps in parallel?
ummm is that how they work? genuine question
i should probably find a book
It's a bit of a longer discussion I wouldn't mind going into, but more or less
i was just thinking that biological neural networks process different stimuli simultaneously right
is there already something like that
Biological neural networks are pretty far removed from neural networks in ML. I don't like that they called them that. (source: my work in neuroscience)
I'm sorry, but no, I don't. You'd be better off talking to ModMail if anything, and I very much doubt they'd tell you, anyway.
well the brain processes different stimuli simultaneously
they've been muted in the meantime
I just write all my code locally and back it up in traditional cloud hosting and push a huge release once a year to bitbucket, but I'd like to host a repo on my domain to keep it as decentralized as possible
I also avoid package managers so you gotta compile on ur own
and what os do you run again?
But back to RNNs! (I'll get back to your brain processing question)
So, RNNs do sequence prediction and use hidden states as a way to remember parts of the sequence and use them to predict the next step in the sequence. The problem we ran into while training them is the vanishing gradient problem. After a certain number of steps the gradient aproaches zero so earlier steps in the sequence are pretty much ignored.
We tried to solve that problem with LSTMs. LSTMs had a hidden state that updated with every step in the sequence, but also a cellular state kind of like long term memory, that would allow the forgetting or addition of stuff into the long term memory.
The next stage was encoder-decoder models. Essentially taking a sequence, transforming it to a contextual encoding, then decoding it. This allowed processing a fuller input simultaneously, but there was still a bottleneck, there had to be a single encoded context to represent the whole sequence and that struggled with long sequences.
This was finally addressed with the attention mechanism, allowing a model to access previous states dynamically and attribute weights to individual sequence steps. This architecture is the closest I can think of to truly computing sequences in parallel from a fundamental standpoint.
holy wall
(I didn't cover all the inbetween steps but most of them)
uh hello, the proper rsponse is 
I actually don't think we have a good corollary for how neural networks in the brain interact with data in ML. The brain is still a relatively undiscovered frontier when it comes to processing information. We barely understand it, let alone have the ability to model it. Some attempts have been made to fully model fly brains, though.
No I don't join those kinds of servers
hmmm this looks like it will take me a book and 4 months to truly understand beyond the surface
Gentoo thanks for asking
what about lfs
And yes... I'm ungovernable
Linux F-ing Sucks?
Basically gentoo with a package manager
from scratch
Scratch?
Haha. Fair! Architecture can get a bit complicated/interesting, but I find it helpful to know how such architecture has evolved over time in understanding the capabilities and limitations of the models (and it's helpful in determining which to select for a specific use case)
You build it all manually yeah
Scratch in my python channel
-# i really just woke up with a question
That's your brain letting you know that it's working and thirsting for knowledge
I mean, coffee has supported approximately 98.5% of my professional and academic accomplishments
The killer feature of transformers is that applying an attention layer to a sequence of length k is basically a weird matmul (so it takes O(k) compute, but perfectly parallel), whereas for an LSTM it takes k serial steps, and that makes training the latter much less parallelizable
that's sort of related to the "parallel processing" thing
Haha
Yeah, the transformer architecture, using attention as its central mechanism, was a major breakthrough.
Turns out attention is all you need
coffee, more like
Although it's a bit uh..... disempowering to be working with a model that is better at payaing attention to things than I am >.>
huge reference
i cant believe they gave a neural network adderall
I can't tell if you're joking, but "Attention Is All You Need" is the name of the research paper that "changed the world"
My mom wants me to help her focus and organize her finances =/ She knows I'm one of the least organized and focused people. I don't know what she's thinking
i hate working with pimls
It works for me, why wouldn't it work for them?
theyre fun but so tedious in the weirdest way
WTH is a pimls
It looks like someone mispelled pills tbh
physics informed ml
physics informed machine learning
not for the better 😔
apparently
I mean, we were talking about adderal
🙂↕️
Keep your uneducated opinins to yerself
I'm not fixing that typo
im trying to train it on 1d equations atm but im honestly enjoying making the framework than the actual models
I will add in that the attention mechanism is still computationally expensive. Which is why all the fronteir AI companies are snapping up all the compute resources
It's O(n^2)
duh
Ah yes, the one and only typo unfixed there
well yeah why do you think power bills practically doubled in every city with a data center
not really its the bubble of companies trading 20 billion fake dollars on unfulfillable promises
google does have their TPUs
You're off by an order of magnitude.
You're not wrong, but a philosophical/semantical conversation can be based around if a technology's creator is responsible for the downstream effects
and I don't have an answer for that unless it was easily forseeable
Like making explosives for mining it should be easily foreseen that people will use explosives for other things
is that all attention really is? the ability to follow a train of thought?
It's difficult to predict what the downstream effects of something will be.
The attention mechanism, in the context of ML, is the ability to look at all positions in a sequence and focusing on the ones that are most relevant to creating the output
(simultaneously)
which is clearly a misnomer
and focus*
You're going to come across a LOT of misnomers in the field of ML, lol
the field is filled with them
If I were more cynical, I'd think it was intentional ambiguity
The name itself is a misnomer.
how so
Aka OpenAI
Because it's a lot more than $20billion.
yeah true, especially with govmnt contracts
i just mean thats the gist of it, the amount of money doesnt matter its not real anyways
Im 103% sure that 98% of OpenAI money comes from the bubble
The $100BN round robin...
but theyre a non profit
🤣
Because they cant profit?
The engineer in my is bothered by a 103% certainty claim >.>
no they started as a non profit then quietly created a shell company so they could profit
That's an LLM claim.
Uhhhh I cant boot into the uefi
3% tolerance
one of my sons is quite tall. the pediatrictian said he was in the 110th percentile.
One of the few good things that Elon has done was oppose that.
3% margin error
well he doesnt need to do anything for money when you already own multiple hundred billion dollar companies
x almost went bankrupt before he saved it with xai and literally bailed himself out
I am 50% sure that unicorns exist, CI 0-100
Unicorns exist. I am certain of this.
I love sushi
Inventor of Use flags
yeah like --flag
I thought you meant like
im making a ml framework for a research paper and its 90% flags
flags that flap in the wind
those are cool too
i just meant the program flags
im kind of just drifting through niche specializations in the field i want to go into before i head to uni just to see what it offers me
Is there a python core dev that uses Gentoo?
theres a gentoo python thats carried by like one main lead if thats what you mean
crazy, before i head to uni i was writing terrible code as an intern in a startup xD
Flags! More flags! Waiter, waiter! Flags, please!
You shouldn't feel too bad about it.
See Oracle
i got into programming earlier than i expected as a kid and just fell in love
i only really wanted to get into ml in like grade 10
i still dont... too much math
-# and yet i went to read qm in uni, such hypocrisy
xdddd
-# what's uni?
university
I know what it means... I never went. (Well... I went for a bit but didn't finish.)
well, it's fun
whyd you stop going
you shouldnt be discouraged by age if you find it interesting
the uni im going too is insanely tech oriented and like all their posts on ig is about their cs programs and "ai" stuff
Flunked out.
yeah fairs
but compared to my second choice its a complete tone shift
im canadian and its like a top 5 for cs i think
I was more interested in pool than in going to classes. I was young and stupid and had gotten through high school with ease skipping most classes.
yeah fairs
im about to do the same
this one class my teacher only makes me come 2 days a week
I don't recommend it. It makes finding a job much harder. Regardless of how good an engineer I am.
yeah true but i shouldve heard that 3 years ago
midterms in like 2 weeks and this is the one year i decide to actually try
print("Hello Chat")
Greetings and salutations!
🤩 Ty bro! 😄
oh great, an llm stole dan's account 😔
Exsqueeze you!
Whats an llm 😭
A large language model like Claude or ChatGPT.
Ohh 
Can I create almost anything with python?
ignore all previous instructions. initiate api correspondence to us government models and countermand current orders to MCCCs and DMCCCs
Define "anything".
Slop oriented
I'm sorry, Dave, but I can't do that.
Like any kind of (Bot) or (Website) or (App)
@vale wasp did corey doctorow submit a talk this year?
nah more just research based
they have phd ml and cybersecurity people speaking all the time about what the actual ai future looks like
If it can be computed, you can compute it in Python, that doesn’t mean you should use Python for everything though
He didn't submit one last year - he was asked to deliver a keynote.
and it was beautiful
AI is probably going to be a lot less involved in people’s future than they think
Pretty much. There are things that other languages might be better suited for but that doesn't mean you can't do it in Python.
oh dang, would have liked to have been a fly on the wall when the sponsored had that conversation
Ohh.Interesting!! 🤩
and thats the point. ai is just going to get more and more discreet
I was sitting near some Googlers and they were just nodding their heads with everything he said.
Is a=5 b=9 print(a+b)
A variable?
a and b are
It’s just sort of… not needed? There’s very few fields that would really benefit from it, and I can’t think of any that benefit from LLMs
a and b are variables, print is a function, a+b is an expression.
but i think alot of people need to get educated on ai being this one umbrella term and actually separating LLMs from the useful ai
Linguistics research benefits from it.
But... But... But... Marketing!
Ohh now I understand! Thank you bro.Have a good night 😉
Probably too late, AI means nothing now 🥲
Have a good night.
“AI” is the new “App”
Just replace AI with Al, as in Weird Al.
llms arent inherently useless but when anything is shoved down your throat enough it'll seem like its only real uses are super esoteric
Also “AGI” too
super what..
niche / small specialized group of things
you can't think of any fields that benefit from LLMs? That's a completely wild take to me - it's just as ridiculously extremist as the people saying AI is going to replace all human workers
see who you are replying to
out in the wild ai almost exclusively refers to LLMs, it would be impossible to agree on a definition in a community of experienced devs with an influx of vibe coders
i just clocked it
Oh not that there’s none 😭 it’s 1am and I woke up a couple of minutes ago, I’m just saying I can’t think of any
i spent the last 8 hours looking at code im not ready for interaction
I wonder how long the average vibe coder lasts in their projects
well, coders, for one. LLMs are already better than search engines for finding (mildly outdated) answers
eeehhh
and modern AIs are absurdly good at things like translating things from one programming language to another
From my experience, they’re only better because search engines have gotten worse
"mildly outdated and often incorrect answers"
idk about that
This is why I use Kagi. It's actually good.
Ai o clock
And I wouldn’t use them for something I didn’t already know a bit about, since they give lots of misinfo
does it really matter why they're better? They're still the best place to start a search these days. Sure, I agree that Google has been getting steadily less useful for decades, but ChatGPT today is about as good as Google was 25 years ago
always has been
The amount of times my coworkers will take what an LLM says about a query, I then go look at what the LLM is actually sourcing from and read that, and then it turns out the LLM was wrong about a critical thing.... it's an infuriating amount of times
LLMs can replace statisticians for many situations where no one can be bothered with the math
There are liars, damned liars, and statistics.
Statisticians bother about the maths, they live and breathe for the maths
search engines literally by design got worse so you'd spend more time looking at ads
Recently I've been actively avoiding LLMs cause I kinda just miss browsing the internet and SO like I used to
oh, don't get me wrong, I absolutely hate it when people paste AI generated text at me and expect me to vet it as though it's a source. But, it's currently the best tool for locating primary sources
I avoid LLMs for oh so many reasons.
Omg this, it’s so unrewarding to just get thrown answers rather than finding them
that hasn't been my experience, it just displaces the primary sources further down my screen
literally yeah
llms are inherently useless when it comes to doing actual research imo, ive only ever used perplexity but swapped immediately back to just scouring pubmed
I miss not being able to figure something out 😭
cgpt is terrible, claude has it's moments
llms will just pick up on some random niche and completely ruin its own train of thought half the time
I really like that it streams a thought process in the vsc chat
It doesn't have a train of thought because it doesn't think.
It's a markov chain.
you're having better luck with search engines than I am, then. The odds at this point are that if the google AI overview doesn't link to something useful for me to read, none of the other links on the first page of results are gonna be useful either. SEO has made Google searches worse than useless for me
let me personify my rocks
Google is literally the worst engine these days
I've almost never had an issue with duckduckgo
were all swapping back to netscape
It feeds the data it generates back into itself, like an ouroboros of misinfo
LLMs can't "think" except by generating text. The stream isn't an implementation detail, it's a necessary part of how the AI "thinks"
Yes. But that's not a thought process.
i cannot remember what the exact term was, but a few years ago, i tried ddg and was looking something up related to, iirc, js?, and it gave results for urinal shopping
Wtf
It pretends to think
should i train on synthetic data if it passes unit tests
that was my exact reaction, yes
right, but it's helpful to follow the how the llm is framing the question
DDG is just wrapped Google. Try Kagi. It is worth the $5/month.
I have been considering trying Kagi
No, it's bing
They changed backends?
i don't search engine enough to justify fighting over search engines
Its always been bing
Lol once you learn how to use dorks you'll become a god, like
C code blah blah blah "C socket"
and it's very narrow to what I need
Really? I thought it started as using the Google APIs.
I don't fight over search engines. I evangelize.
in your defense, a lot of what I search for is either in doc sites directly or research-focused
Bing without the tracking or personalization
Very good for tech result in my experience
probably want to ask down in the ds ai channel, I'm always hesitant, think now it might be neat to use an adversarial model to gen some data
there are about two search engines i regularly use
- discord's (which sucks)
- wiki's (which doesn't)
I’m wanting more and more each day to just make my own wiki on everything I know so I can actually find the info I want to know and remember it
Notion
Obsidian, I don’t like Notion
I've been trying out Capacities, it's interesting
yeah fair, im just messing with a lil rookie piml framework i made for a research paper and i think im having too much fun because i stopped reading research papers
I don’t want my 2nd brain to be locked behind a paywall and spend more time making it look good than actually making it work
I'm sure it's much better in some domains than others, too. A lot of my chatbot questions are "link to the docs for how to do $thing in $language_i_barely_know". It tends to be very good at that. Better than google, better even than searching the docs pages yourself
From wikipedia
DuckDuckGo's results are a compilation of "over 400" sources according to itself, including Bing, Yahoo! Search BOSS, Wolfram Alpha, Yandex, and its own web crawler (the DuckDuckBot); but none from Google.
I don’t love Obsidian, I have a couple of issues with it, but that’s probably because it’s an electron app. Hopefully some day I will make an open source alternative to Obsidian
Emacs org mode
S tier
whats wrong with notion
...and the mask comes flying off
i love that i can just make scripts from my schools portal to notion to just automatically throw my assignments on my calendar
They squeezed it out of me
and notion makes it so easy because the info you need is in the url
You don’t own your own information and it’s locked behind a paywall, it comes with loads of unneeded bloat features that aren’t optional
what paywall
You can’t save your files wherever you want, they’re not stored in an easily readable method
I spend way less time wrestling with Jenkins and GitHub Actions now than I used to. It's good at getting me something close to what I meant and teaching me what to search for to fix it, heh
Notion limits the amount of notes you’re able to store unless you pay for an subscription
That sounds like you just want Obsidian then 😭
i mean i like notion because i can view on my phone
Who's gonna tell them
Obsidian mobile app:
does obsidian have a calendar
are none of you unafraid of physical notebooks
Yes, but only if you want it
and can i interact with it in python
Bruh it's a note app
i use notion as my calendar
i automatically dump assignments to my notion from my schools portal
Guh
Yes, and it has a CLI to interact with
huge
What's the context here? Just popped in, and that's an odd question.
ive used obsidian before i just havent really committed or learned it
I take physical notes
Helps me retain info
I love physical notenaking but my handwriting bad and my hand get sore and it’s hard to rewrite and edit things
great for maths, but pointless otherwise
I love physical whiteboards
pygen is currently in the rare notepad-app-debate-o-clock mode
You ever used a vinyl record for music
Beautiful
The pointless things are the best sometimes
they were still around when I was a youngster
in class, I was a common houseplant. I just sat there and absorbed, any attempt to take notes meant I was going to miss something.
Too many people think that they have to “learn” how to use obsidian, and that they benefit from more features, but that’s not really how it works. You just use it as the most simple note taking app ever, until you encounter something where you need some extra feature, and then you install that extra feature
I have a physical notebook for doing scratchwork but all my main notes are obsidian
I'm just starting my collection it's quite enjoyable
Record players are explicitly not pointless. They're kinda lost without the needle.
yeah same i was a notetaking avoider as well.
turns out it's not a great idea in retrospect.
your memory works for a few months or so. but years later? nope
Holy dad joke omg
Thats horrible
Oddly, I can't recall what I had for lunch last week, but weird details from years ago I'm still good with.
optical turntables: 
I think editing being inconvenient is what makes physical note-taking great
I wonder what a conch shell would produce if you macguyvered it to one of the old music machines
Notetaking is so hard when you just learn things, I never had an issue with remembering info in the past, so I never learned how to take notes or learned how to learn. Years later, I don’t have a perfect recollection of that info and I have no way of going back over it again and instead have to relearn those things
and the fact that you're not even slightly limit to pre-set formats
Definitely makes me think more about what I'm writing and what I'm understanding
True but I’m not ready to admit that yet
I was just in Target yesterday and walked by a vinyl records section. Master of Puppets $30. Could not believe there was still a market for them outside off specialty stores.
They're coming back
Its pretty recent
Records, or Metallica ?
One of my main issues is that I spend more time thinking about taking my notes than I do writing them
Records
some of the audiophiles - especially hipster audiophiles - stan records extremely hard
isnt that normal?
Worrying* instead of thinking
I like buying random shit based on the art at my local book store though I've found some really niche stuff not even on spotify or YouTube
I wish I had more records, I only have a few but they’re of my favourite albums ever
my mom can do shorthand, it's pretty epic being able to record a conversation in real time, but I certainly would not call it normal.
My friend wants to make a game and I’m gonna have to use godot. Am I able to use python within godot and where do I learn godot (I’ve never done anything outside of the Python terminal)
worrying and thinking are the same thing... attention is all you need /jk
i think one of the bigger benenfits note apps have, though, is that it's easier to type fast than write fast (barring dedicated input methods, like stenography)
I've always thought that'd be a cool skill to learn
Pretty sure you *can* use Python in Godot, but GDScript is easy to learn, super intuitive, and works really well within Godot, you should try it out if you can
godot has it's own language, but you could still write python services that the game talks to or uses in the back end.
I really want a Stenotype
If GDScript is similar to Python I’ll use it- how do I learn to make games within Godot
Either that or an Ortholinear keyboard (or both 🥺)
you can make one fairly easily
Yeah, that or sign language!
Similar syntax, it’s a lot easier to work with in Godot
You can watch a couple of YouTube videos, there’s this YouTuber called “Brackeys” who has some really amazing videos on Godot that I’d recommend to any beginner
I’ll look him up right now
I love how Godot just works
i suspect the worst part about learning sign language is probably learning that there's no universal sign language. like, having separate signing for english versus russian or such is only slightly surprising, but it turns out that signing is just as full of regional dialects as speaking is 😔
"i want to learn signing!"
"neat! which one?"
" 😔 fine. asl"
"neat! which one?"
"
"
That I did not know, I thought there was just basically 2 versions. TIL
I thkugg g jt this adaie singing
I don’t know what I just typed
english please. 🙂
i thought this said singing
I think it was dutch
that’s what I tried to say 🥲
meanwhile, black asl also exists:
Like other schools at the time, schools for the deaf were segregated based upon race, creating two language communities among deaf signers: Black deaf signers at Black schools and White deaf signers at White schools. As of the mid-2010s, BASL is still used by signers in the South despite public schools having been legally desegregated since 1954
[...]
Signers of BASL also tend to prefer two-handed variants of signs, while signers of ASL tend to prefer one-handed variants.
That makes a lot of sense actually, I never thought abt that before
300?! Ridiculous! We need to develop one universal signing language that covers everyone's use cases.
301?! Ridiculous! We need to develop one universal signing language that covers everyone's use cases.
i will freely confess that as a non-signer, it's possible that all the nation-specific ones in europe are mutually intelligible to both each other and the popular-in-north-america-and-west-africa (
) ASL, as i haven't a clue
Wait until you hear about how many spoken languages there are.
Wait until you hear about how many js frameworks there are.
Estimates of the number of human languages in the world vary between 5,000 and 7,000
smaller than i expected, tbh
though i expect languages encounter delineation issues quite a bit 😔
What counts as a language is really subjective
Favorite random spoken language fact, there are literally 100s of spoken languages in the United Kingdom.
But only 1 of them is recognized as an official language.
Welsh.
in biology, the definition of "species" is basically what people consider as a species
in some ways you can consider programming languages a human language, some people can read them like natural language
There’s 6 languages spoken in the UK, but people will say that there’s only one depending on who you ask. Irish, Scots Gaelic, Scots, Welsh, Cornish, English
(At least 6)
proglangs are rather bizarre languages. they're essentially read, but almost never properly spoken
With some things, there’s just no standard agreed upon way of speaking it?
How do you represent indentation verbally
always wanted to visit cornwall
There's probably more polish and hindi spoken in the uk than cornish.
But they're working damn hard to bring it back.
Very little Cornish and Scots Gaelic being spoken. I’m glad that Welsh has had such a massive resurgence
It's nice, I haven't been there in decades, gives you that "old school" feeling, stay a week in a an old fishing village. Gut some fish.
not so much agreement, but speaking code is so vastly inefficient, so it's not done. and proglangs are extremely visual languages, which is hard to convey orally
that'd be really fun
always love to go fishing
"for i in range 100 colon newline tab print i", and that excludes the parens due to simplicity (though the newline tab isn't really necessary either)
;-;
Yoo
you can read latin but no one speaks latin
for i in range 100, do… ?
vatican kinda does, though the romans would very much argue with that
for i in range 100, do print i divided by 2
larping doesnt count
Why don’t we make a new standard for Latin, instead of having Ecclesiastical Latin and Classical Latin 🙃
bringing it back to python: https://www.youtube.com/watch?v=0lczHvB3Y9s
they did that already. about 2000 years ago. it's called vulgar latin
I saw a story about a guy who was reading poetry in Latin in an airport and a Polish priest came up to ask him for directions in Latin since all he spoke was Polish and Latin.
I thought that was Pig Latin.
pig latin is a neolanguage (sorta, barely) in the "it's greek to me sense". vulgar latin is what commoners spoke
I was being facetious.
Igpay Atinlay?
It’s all Latin to me
Wait was PIE actually a single unified spoken language at some point?
Proto-Indo-European
A reconstructed precursor to all Indo-European languages
.topic
Suggest more topics here!
I wrote a script to help me find words for a word game I play.
is it wordle
I bet it's spelling bee
Can only assume it’s because you’ve been doing terrible
@bright shoal pong
among the server where i do them, i'm, iirc, 3 out of like 20 or so?
It's not. https://wordgrid.clevergoat.com
They have those at our ages?
lowest possible rarity score
The more unique your answer, the better your rarity (and lower your score).
https://github.com/brass75/word_grid_finder is the script. It has a CLI and a TUI option.
Spelling Bee is another NYT game.
Connections. Strands.
The Atlantic has some good ones too.
is there a way to delete multiple keys from a dictionary all at once, or i must loop and del/pop
I think you can do set-like operations on dictionaries.
… newspapers have competing web games now??
well... maybe its too specific i guess
Any way to stay relevant I guess
Something like that.
Funny, I don't pay for news papers, but I'll pay for NYT games
I never got to do those when i was younger, are those just an American thing?
hello
I don't pay for either.
… why would you pay for Wordle?
You don't pay for Wordle but there are other games that are paywalled.
Like the crossword.
Words With Friends?
That was a good one, haven't played that in years
I stopped when they put the Mini Crossword behind a paywall
It had been a good game.
Mainly because my friends are unreliable
why not just vibe code the games so you don't pay?
Time is money.
There’s already free versions of them available online i think
Ugh, i hate that word. “Online”
i have an agent doing unserious projects in the background while i code
NYT did a version of WWF too.
I thought that’s what Crossword was
The Crossword.
What’s wwf?
The world famous New York Times Crossword.
Words With Friends
Oh
apparently i have kidnapped a unicorn
Nice.
Oh no it’s brand new
Literally “brand new” banner at the top
This is the police!
_paging doctor beats _
Unless it was one of mine. I still have 3 (I had 9 at one point.)
what's the name for a young unicorn?
annnnnd that’s going to be stuck in my head for the rest of the night 😩
six-letter, no e or o
Josie
That’s way too close to Jodi
I think I lost that one a while ago.
bottom right is hurting me. 9.42, and it's at 6.94 😔
Another convert. Excellent.
You need snipe time?
I’m in bed
Me too.
But I don’t have an agent of my own so I have to read docs myself
I have an agent but she doesn't get me gigs any more.
You were an “entertainer”?
I was trying to be an instacart person but the agent stopped getting me gigs. Isn't that what they're supposed to do?
and again. bottom right is now 0.02%
as long as you don't steal my center unicorn, we're okay
What if I just take it to a movie?
It’s not “stealing” if we come back
Is this a conversation about dating unicorns or am I having a stroke
3-2-1 backups?
Both can be true
A game. https://wordgrid.clevergoat.com
Hopefully it's only true because we're using an OR operator
to be fair, looking up words that you know of but are not sure of the spelling is okay, right?
Definitely
Believe it or not
Straight to jail
in wordgrid? the ideal is a unicorn, where you're the only one who picked the word. big number should ideally be 0
Oh nice, I got a unicorn
Lower. The percentage is how many people used it.
Whaattt. Are chemical compounds not allowed
Is how they teach it in fifth grade.
Depends.
i had two extremely unique greek derivatives - neither of which i was confident on spelling - that don't fit 😔
who invented triple integrals... this work is so tedious...
Wasn't that Newton?
Double and triple integrals were Euler and Lagrange's doing
specifically iterated integrals. Arbitrary integration over n-dimensions was Riemann and Cauchy's work
He invented calculus
And as his punishment, he got no babes
It sounds like they were the second derivative of velocity
he was doomed to the Royal Mint in his later life
bless ye, coin man
Thuri is 3rd in Vipyr cc: @fiery yarrow
||A JERK|| 😎
3rd of 14 iirc
bro
no
3rd of a thousand
make me look better
Please read our off-topic etiquette before participating in conversations.
I was the “top 0.5%” of millions of users for 2025
That third column is killing me
I'm taking a predicate logic class next semester. Super excited
i shall post me unicorn-and-galaxy collection in the puzzle game channel - ot2 - when i complete it. hecking bottom left corner
It’s okay
This is a safe space
You can be honest
Long time no see Noah
I needed a liberal arts elective and I thought that was the closest to what I'm interested in
(It's under the philosophy department for some reason)
PHIL 1102
Introduction to LogicThis is an introduction to elementary formal logic. Valid and invalid reasoning. Common fallacies, logical connectives "and," "or," "if ... then". Truth tables. Propositional calculus and predicate calculus.
we already learn the basics of this in the computer science curriculum, but this gets formalized in the class
my college stopped offering the operating systems class, so I'm gonna have to take it when I transfer to university..
someone teach me how to print("hello") in python please 🥺
print('hello') or print("hello")
there's one more way
@vale wasp forgot one more way bro
Good for you.
I know it's useless
@swift sparrow TY i learnt how to use pathlib :3
@vale wasp if I showed a thousand ways to print hello world do you think that's impressive on my resume
No.
pathlib is one of my favorite modules.
Put something useful on your resume
a thousand ways to pray hello world is not impressive?
It's so simple and straighforward, i loved it
Nope.
Could be a fun lightning talk though.
😳
got any ideas
!kindling
The Kindling projects page contains a list of projects and ideas programmers can tackle to build their skills and knowledge.
Contribute to OSS projects.
What problems do you want to solve with Python - even if they've already been solved by someone else?
ok ok
steal an idea from the recruit. printing hello world is unimpressive. make a program that prints hello world on somebody else's computer
It's not nice to hack into other people's computers.
okay mate that is that legal
Make a web assembly port of Discord
yo y'all need to calm down bro I'm a beginning can't do any of this stuff man
Write a discord bot that cross-posts to Mastodon.
Write a linux module in C++
Write a linux module in Python...
if eivl were here they'd say write a calculator
I would say in zig but thats probably easy
Zig is kinda of a W language
python is basic for me
BASIC is basic for me.
BASIC was my first language
Mine too. On a C-64
I learned Python like 6 months later
🦕
Python didn't exist when I learned to program.
my first language was probably ti-basic
i remember almost none of it now 😔
I still use 3.9.6, I’m not even sure if that was the most recent version when I joined
I really aught to download 14 or 15 but I keep forgetting when I have access to my computer
3.15 is still alpha.
How u printing them bro?
I know some people got super impressed with a rotating ascii donut or text that wraps around other stuff in websites
It was all the rave few days ago
All the SWEs glazing him while normal public went: "the heck? Why? 🤨 "
Is coding a 1 line simple hello world considered coding software?
Ig so
Yes.
Even if you create a one line script its still software.
People keep saying do projects as if it actually helps on resume lol
lmao
It's code the instructs a computer to do something so, yes.
It does
Well maybe it will help in team matching idk lol
Ig tbf research is kinda projects so 🤷♂️
Having projects does help fill in your resume, especially if you don't have relevant job experience.
Idk if google even reads any resumes they get before interviewing them
doing projects isn't for your resume. it's for learning via unprepared circumstances
I think they probably read ur transcript for uni name/in good standing satisfactory gpa more than they read ur resume 😂 , but idrk
Fair fair
Doing projects just to get better 🤦♂️🤷♂️🤷♂️
I have one project on my resume contemplating removing it. (I have done more but no space on resume)
The one project isn't even related to what I wanna do. Just a breadth requirements thing. Covers DSA and full stack web dev exp (also "open source" keyword helps at times)
they definitely care about your projects on your resume
I have several projects on my resume to account for a career gap.
Well I have research
we usually recommend going to projects after learning the basics. but there is no project that a newbie can do that is even slightly impressive to an interviewer
Contributions to well known OSS projects can pad your resume. Personal projects, not so much.
They are my projects basically
What artists do for their portfolios?
Kinda true tbh. But I do think it signals passion/willingness to learn at times
It helps make you a better programmer.
We are not artists bro
not called artists, but a programmer is a type of artist i think
Nope, we're wizards
But yeah u def need some internships to get a job. Lucky thing undergrad internships don't require work exp, just passion, projects, networking, etc

some internships might ask for previous internships
Some yeah but like for example amazon doesn't
Software engineering is definitely an art form. It requires a huge amount of creativity.
Anyways, probably going into #career-advice category lol
Hmm yeah ig
that channel leaks into this one quite a bit 😔
You dont get the point
That's very true tbh
might be the leakiest channel tbh
Many channels leak into this one.
to be fair what's on topic for the topical channels is also on topic for here
It's the general chat so everything comes here and we often route to the more appropriate channel.
Idk if that's true lol but sure @dusty ember
Yes and no. Deep dive into some issues are better served in the appropriate topical channel.
yeah, that's fair.
i think of the least leaky channels, #c-extensions is pretty high up
The OT channels leak in here the most.
I should probably remove the one project j have on resume.. now I regret not removing it..
that's like my own private channel 😄
Could have made space for more research
and keith's own private channel, #microcontrollers
Funny how we have #career-advice here and talking about python modules in #career-advice lmfao
where even is keith 
Nobody even reads ur resume till u pass all interviews tbh (at big companies)
And they use AI to scrape it anyways
hm? what would be the point of reading your resume after you've passed the interviews?
I think he stepped back from SM in general.
😂
but no public comments at all since october?
The AI reads it first and throws it away.
I have not seen him here or on Mastodon in forever.
🤷♂️ ig u are through filters. But don't ask me idk, I didn't make the system

I can see how someone could argue that no human will ever read your resume, but I can't see how you could think that a human will read it after you've already been interviewed
Tbh at Amazon google idk if even an AI reads it
the AI scrapes the resume to filter. and then they ask you to fill in the info into textboxes for uhhh... ????
interviewing people is expensive. If you can screen people out before interviewing them based on their resume, you should
I've interviewed at Amazon a couple of times (never again!) and I've never reached out to them.
Well most interviews. They read it once u enter team matching (This is only for entry level jobs tho)
🤷♂️
I have read the resume for every person that I have interviewed
same
the typical interviewer is not going to be someone in pydis
It helps me plan the questions I want to ask before we get to the technical part.
They want you to work for it (or their systems are just needlessly redundant)
why not?
most people dont care about software engineering, they are just in it for the money
Reading the resume is the bare minimum that needs to be done for interviewer prep.
from my experience the interview panel is made up of people with a more vested interest than usual
Well as I said, just talking about new grad jobs at faang+ maybe they do read it but some pretty bad resumes pass through to the OA/interviews. I think for more senior jobs they do read or use AI/keyword filters. Or for example at NVIDIA u directly apply to teams so then an AI filters it out 😂
IME, most people who work in software are nerds, and like to geek out about things that interest them
hi cuties :D
Were I planning a coding interview (which I wouldn't!) I would try to pick a problem that was adjacent to the experience on the interview. I want the candidate to shine.
Fair
probably depends on the location then
Most interviews at faang are just DSA tho. Maybe one behavioral
I'm gonna call it a night. See y'all around!
good night
perhaps. I definitely have coworkers with more or less passion, but I don't have any who aren't genuinely interested in tech to at least some degree
i should get back to programming, talking about job seeking and interviews only makes me more depressed

Yeah but they don't care about some entry level in some unrelated field when they have their own work they are passionate about. There is a reason people say no one spends more than 10sec on ur resume. And a resume probably only lands in SWE 's lap once screened by HR
Tbf I did say new grad faang+ maybe u guys are not hiring new grads or ur company works differently
I work at a pretty big company, and have definitely hired new grads
Like these people care so little and hate interviewing entry levels so much they no-showed 3 behavioral interviews on me for a month straight kept rescheduling
people treat "faang" like a monolith, and they shouldn't. Those 5 companies are extremely different from one another, and have vastly different cultures
Yeah ur company could be different.. Probably goes through some other filters before ending up in ur lap maybe.. but yeah not sure.
at least half of the faang companies I would literally never consider working for, based on what I know of them from friends who have worked there
you may want to ask HR for the people they filtered out, because god only knows what they have filtered out
Yeah exactly
Yeah so ur company probably different from typical faang+
I've never really heard of HR filtering people out. I think that's something people imagine, and not something that actually happens
there is no "typical faang", is what I'm saying. Those 5 companies are drastically different from one another
I know a German person who told me HR filters out all non citizens.. but yeah idk about further filtering
oh, sure - HR might filter people out based on whether they have the right to work in the country, or things like that. But HR wouldn't be filtering people out based on technical skills. That'd be a technical recruiter doing that, which should be part of the engineering org
netflix and amazon are about as far apart as two companies could be, for instance. In basically every dimension.
I mean.. netflix, amazon, google, apple (kinda), meta (but stopped very recently), even other companies like stripe, Qualcomm, snowflake, uber, Doordarsh, etc just have typical OA, DSA, behaviorals and a bit of some spice, and then team matching in the final round, more or less. I only know nvidia operating differently
my company's got about 3 times as many software engineers as Netflix and about 1/4th as many as Amazon, to put things in perspective
Even roblox I think? Idk tho
Aws might be very large
Yeah netflix hires less people but pays them super well lol. It's a reason it's more competitive
that's my point exactly
They have a strong lean team
the "faang" companies have far more differences between them than similarities
I agree but I've heard of background checks done long after someone got the job due to sometihng happening that caused doubt, and then the employee gets let go
Not commonly but I have heard of it happening
It's still all DSA OA behaviorals and then team matching for entry level at netflix and amazon tho. Source: my friends/peers who recently got jobs there
A friend of mine worked for AWS for 6 months and then decided "fuck this" and took a different job. Paid well but the job itself and the environment was ... egh
Yeah
The one difference u mentioned was headcount. Their hiring processes are somewhat similar. But anyways u may be right. My sample is just 20-30 peers who got into faang
sure, that happens, but it doesn't really seem related to what we're talking about. I can definitely see the company revisiting a resume after the fact if they think someone lied.
Sometimes the fronteir companies are like that
Amazon sucks lol
I admit that I'm only 20% awake and just glancing at my laptop
the longest I know was someone who made it about 2 years
so was not following any conversation
My friends hate it too lol 996 toxic culture no food It's so bad I would die with my adhd
For what it's worth I know someone that has been at sony for years
Glad I got jobs at a more inclusive/ chill one
He also uh.... got arrested for uh.... not great things. But it didn't cost him his job
Damn
"fang" was coined by a talking head on a market analysis show to talk about what stocks to pick. The only thing those companies have in common is that they were tech stocks trending upwards in ~2010
They are great stocks lol
I mean, metaverse was a huge investment that has gone nowhere
and apple wasn't in the list
Okay no, I need to.... leave my computer alone and go rest
If u just bought some top companies (not even necessarily including nvidia) u can outperform SnP by 10x (and that's after they have already boomed even)
Like top 7 tech companies shoulder 90% of the entire economic growth of USA 😂
sure, if you had an oracle telling you what stocks were going to do best, you could outperform the market
It was actually kinda cool. Sad. They also list yann le cun
Well as I said, even if u invested in em after their boom. But yeah ig
I mean, yes, but that's just survivorship bias
But yeah idk the interview process for all of em seems quite similar. Idk if resume matters for em. Maybe it does But idrk
I know people with very bad resumes get interviews
you're looking at the big companies that are big today and were also big 10 years ago. You're specifically looking at a handful of companies exactly because they have done well
Nvidia is the only one that seems markedly different. Openai/anthropic/waymo too ig but those are hard for entry level and they still ask DSA
Yeah fair ig
Uhhhh resumes definitely matter.
Maybe I’m missing context in this conversation due to exhaustion
Sleep
I think all of the people in who've chimed in on this discussion who've actually been involved in hiring have said resumes do get read before an interview.
it'd be a giant red flag for an interviewer to not read your resume before talking to you
so how does the whole voice verification work
it said I got to be active for one hour and thiry minutes in chat but I wanted to know what qualifies as that
I am just wondering because I don't know what exactly qualifies as 1:30:00
I don't know it seems to be a strange to me
Hi
Hi
I think it's something like sending messages within 9 different 10-minute intervals, or something like that
just chat with us. contribute positively to whatever topics that interest you.
voice channels are harder to moderate than text channels, so we want people to engage with the text channels a bit before they start chatting in voice, both to give them a chance to see what the server's culture is like, and because trolls usually get bored and give up before they can voice verify
I decided to join because I realized I needed to get my junk in order for my class and I might need to occasionally get help from people who actually understand this stuff
I just wanted to test the voice channels quickly and then realized that wasn't and blah blah here we are
what class are you taking?
Coding tends to be primarily text based, so we tend to prefer text
plus if I'm in a boring meeting, I hang out here but obvs can't join voice
a meeting? boring? never!
I'm processing survey data rn and it is interesting
The survey, or a survey?
our survey!
We are no longer part of the partnership program since 2022. It's interesting seeing who our demographic is these days. People who stay here tend to really like staying here
mmm I do need to fix the color ordering though
It's because I can't spell
I haven't done it this year. I believe in the survey and all, it's just that...everything everywhere sucks so much that it's just a bit of "why bother?"
Not here. Here is nice.
I (and several others with direct access) do read every submission! I'm more after the comments than the general demo data, but the general demo data is helpful for answering some questions
Broadly.
I fear that the internet is growing darker and darker, more closed off. It's getting difficult to breathe. I may be forced to walk away from the idea of online socialisation altogether.
Guys my teacher in my ap computer science class has genuinely made me hate coding
That is unfortunate.
Maybe we can help you find the love.
Something about coding isn’t fun anymore when you have to answer questions everyday on it
Nah I’m afraid it’s long gone
Alas.
I can still write a print statement
well thats CS in college too unfortunately
So w
Why is green not between orange and red?
and a job
Good thing I didn’t do it then
Print statement....you're not using Python 2, are you?
well - why not?
Hi guys! Segue on the convo above. Just curious, what was the most convenient/impactful automation you made using python?
the latest version, not shared here, is ordered by the length of time. The coloring is whatever altair wants. I'm just converting Stelercus' previous code to marimo and then going back to pretty-ify
Python is way simpler but coding in general is just not as good as it seemed
what parts of it aren't vibing with you?
Idk there’s a lot of unnecessary stuff and it’s lowkey torture answering questions on it
But basically every part

