« first day (4357 days earlier)      last day (484 days later) » 

12:05 AM
@emanresuA It appears I have already offered a bounty on the cops thread, do you have another post you'd like rewarded?
 
@emanresuA My final writeup for this is almost a third of the size limit lol
 
300 downvotes, yikes
 
I was wondering how you hadn't earned any rep all year, then I realised...
 
12:49 AM
Damn, Newly Featured Posts is slow today
 
1:08 AM
@ThomasWard hey wait you're not native to this room! :p
 
@Ginger I exist EVERYWHERE! I exist NOWHERE! I AM THE CHAOS! FEAR ME!
may or may not have had a few drinks
@user ye Spectre for example has some HT related vulns
but mitigated or not vulnerable is fine
 
@ThomasWard oh I remember you, you're the guy who chastised me for having my life's work on two un-backed-up MicroSD cards :b
 
@ThomasWard but are you the one who knocks? :p
 
@lyxal I bet he thinks he's a [Big Shot]
 
1:46 AM
norton thinks my http server is a security risk... with a high threat level
 
watch out... you might steal data from yourself!
and then you'll have to ransom yourself to get your stolen data back from yourself
and then you'll have to report yourself to the FBI for cyber-stealing from yourself. Honestly the best play here is to just shut down your server entirely.
 
 
2 hours later…
3:41 AM
@Simd Thank you so much for pointing out my chip supported 64 bit lol
I'm installing openSUSE right now, didn't give me any errors and it's halfway done installing so I'm assuming it works
 
I assume it's not pronounced "open sussy"?
 
If it doesn't work I will forever hate you for making me wipe my beautiful Gentoo setup
@emanresuA I think it's soo-zuh
 
Aww.....
 
Here, have a video with fake Justin Timberlake pronouncing it the right way
 
4:23 AM
@user my pleasure. Please let me know how it goes!
 
5:17 AM
19
Q: Scan a ragged list

Wheat Wizardhgl has a "scan" function called sc. What it does in general is a little bit abstract, so we will just talk about one specific way you can use it. If we want to take a list of integers and get the partial sums of every prefix we can scan with + for sums and 0 as a starting value: ghci> sc (+) 0 ...

 
6:03 AM
Hello
 
 
2 hours later…
8:10 AM
0
A: Sandbox for Proposed Challenges

alephalphaIntersection area of two rotated rectangles code-golf math geometry Given two rectangles, which are possibly not in the orthogonal direction, find the area of their intersection. Input You may take the rectangles as input in one of the following ways: The coordinates of the four vertices of th...

 
8:39 AM
0
Q: For Haskell programmers, some curry advice

occasionalrifleThere are many similarities between Haskell and Curry, therefore many advice for one language also apply to the other. As a result, I'd want to collect some Curry code-golfing advice in this question that would be helpful to someone who has gone through our Haskell advice in great detail. This wo...

 
@NewPosts
That link might be spam
 
yep it looks to literally just be a copy paste of codegolf.stackexchange.com/questions/245961/… with a spam link at the end
and a different title
 
wow -10 score in 8 minutes
 
9:23 AM
@mousetail I've seen worse
 
9:43 AM
@cairdcoinheringaahing can't believe Jelly has tax exempt status and I have to pay code golf taxes smh :p
 
Don't worry, Vyxal Inc. is technically based in Code Review.
 
@WheatWizard Elm Lotm?
 
Imagine if there was a progressive tax on code base size…
 
I wish. I could finally have the time to delete some of the huge amount of unused code at my work
 
10:00 AM
@emanresuA that implies that code review is the Cayman Islands of SE :p
 
10:30 AM
hello all!
 
 
2 hours later…
12:29 PM
@lyxal that used to be Windows Phone before Stack Overflow nuked it from orbit
 
 
1 hour later…
1:42 PM
0
Q: Maximum of outer product of integer vectors (in linear time)

SebastianMaximum of outer product of integer vectors (in linear time) code-golf math algorithm Our goal is to efficiently find the maximum of a large amount of (redundant) data. We define the outer product of vectors \$A\$ and \$B\$ as a matrix containing the products of all entries of \$A\$ with each ent...

 
2:16 PM
Quick dupe check, there aren't any challenges about the lehmer matrix, right?
 
it looks fun, how did you come across them?
 
I don't even know, I was browsing through random wikipedia articles.
 
that sounds fun in itself
 
2:33 PM
Aren't those about the lehmer sequence?
Actually, scrap the idea, it's a bit too trivial.
 
2:44 PM
@Simd why did you delete your old account?
 
@mousetail It's a personal thing. I really don't like extended records of my activities online
I should have given it all away in bounties first of course
 
That would sound like you'd change accounts every few months or so
 
@UndoneStudios few years
 
anyways, can anyone tell me where exactly the TypeError is occuring in def x(n):t,m,j='','aeiou',0;[t:=t+(m,m.upper())[i<'a'][j%5]and(j:=j+1)if i.lower()in m else(t:=t+i)for i in n];return t
 
3:38 PM
never mind
 
chatgpt says you are doing something illegal :)
@TheThonnu You can upload code with a fixed font
 
def x(n):
    t = ''
    m = 'aeiou'
    j = 0
    for i in n:
        if i.lower() in m:
            t += (m, m.upper())[i < 'a'][j % 5]
            j += 1
        else:
            t += i
    return t
^ that's what ChatGPT says (very golfed :P)
@Simd thanks
 
@TheThonnu np
 
4:03 PM
@Adám But that would unnecessarily penalize codebases that have to be enormous, like the Linux kernel
If only we could quantify the complexity of a codebase
 
@user just wait until the new APL version
 
It's been here since March
 
@user Tax would be shared between every linux user though so it wouldn't be that bad per person
 
Why the users? Wouldn't it only be shared by the core devs?
 
Shared between every contributor at least
 
4:06 PM
This isn't like the real world, where a company might increase prices to make up for getting taxed
 
@mousetail Probably weighted by amount written
 
@user I mean the new APL version of the linux kernel :)
 
Linux kernel will secretly mine crypo on every linux machine to pay for it's tax
 
Oh no
Then our only refuge will be BSD
 
Is the code size of BSD smaller?
 
4:07 PM
:)
 
What about microkernels
 
@mousetail No, but its so obscure no one will bother taxing it :P
 
I like your thinking!
 
I guess it isn't actually obscure
Servers do use it, right?
 
I say we develop golfkernel™
 
4:07 PM
Microkernels seem the only reasonably option, BSD will suddenly become popular if linux becomes expensive
If only in terms of power use
 
@mousetail they have been the future before :)
 
@RadvylfPrograms Written in a golflang and transpiled to C?
 
They have been the future since forever
@user To Rust of course
 
Why would you transpile to Rust?
 
^
Transpilers don't make typos :p
 
4:09 PM
Rust's borrow checker is useful if you're using it directly. Compiling to it makes it useless
@RadvylfPrograms They do when I write them :P
 
Although I guess why transpile to C either, just use LLVM or whatever
 
True
Transpiling to C may be easier for some languages though
 
Do you want manual memory management in your golfing language though?
 
Who said anything about memory management?
Let the garbage pile up :P
 
I think rust-style borrow checker with maybe some kind of auto boxing system might be golfiest
auto clone, auto copy
 
4:13 PM
When the program ends, the memory gets freed. No need to worry until then :P
You could also use the type system to avoid cycles and then use refcounting
 
@mousetail The transpiler'd handle that
 
Closures would be tricky but still doable
 
And Rust isn't garbage collected anyway, you'd still need memory management
 
If it's a microkernel we could just periodically restart modules to free their memory
 
And instead of defragmentation you just completely reinstall the entire system every once in a while
 
4:15 PM
Let's just require that freeing unused memory is a module
The kernel wouldn't worry about it
 
And use something like the Boehm GC?
 
4:27 PM
That or defragment the RAM and inject code into the kernel to fix the pointers so they point to the new locations
Mumble mumble virtual memory mumble
 
There's a stupid but easy-to-implement GC where you only use half your memory at any one point, and when you start running out of space, you stop everything, copy everything over to the other half of your memory, and start again, leaving the dead objects behind in the old half
 
Couldn't you use all of your RAM and use xor-swapping instead of copying
 
CMQ: What is the most cursed approach to garbage collection? (not doing GC at all and letting the garbage pile up doesn't count)
@RadvylfPrograms What's xor-swapping?
 
@user All of your RAM is a linked list; to access the pointer 0x12ab, you follow the linked list to that index. Any time you free up RAM, it's a linked list, who cares
@user Just a way to swap two values without using a temporary value
You'd just copy the second half of RAM into the first, using swaps, then you'd have the first half of RAM on the right, out of order, which you'd reorganize
 
@user Inverse reference counting. If the amount of objects that do not use yours equals the total number of objects, delete it
 
4:37 PM
@RadvylfPrograms (this is actually how I implemented RAM/GC in a brainfuck-based virtual machine I made where performance didn't matter)
 
@user "Scream test" reference counting. Copy the entire state of RAM. Then delete a random object. Then run the code. If it errors, the object was important. Restore RAM and try a different object.
 
That's absolutely beautiful
What about user input that's been passed before it errored, though?
 
Add it to a queue somewhere so it can be multiplexed to the new iteration
If your memory runs out again though you overwrite the snapshot so better hope that object is used in the meantime or it's gone foevever
 
4:52 PM
@RadvylfPrograms smallest possible kernel that can run a Linux-built program, actually not a bad idea :p
 
@user Don't collect the garbage, reallocate it to memory being used by the program
 
@user Easy: When RAM is getting full, write the entire thing to disk, clear RAM, and load in every possible combination of objects in order until the program doesn't crash
Or, when RAM is getting full start writing objects to disk, oldest ones first, and retrieve them if they're requested
 
@Ginger That's just swap though
Already how it works
 
@user As soon as RAM starts to get full, just wipe the entire computer's memory
 
OOH I JUST HAD AN AWFUL IDEA ONE SECOND
Okay hear me out: Paper Swap
When RAM fills up the computer starts printing memory on actual paper using a printer, one RAM page per physical sheet, and when it needs a page you have to put that sheet of paper on the scanner to load it back in
probably using some kind of matrix data representation or smth
 
5:02 PM
What about as punch cards?
 
also, just broke 6k site-wide rep!
 
Congrats
 
@mousetail you can't really print those, and I like how paper pages can work as physical pages
 
True
 
this is a godawful idea and I'm trying to resist implementing it
 
5:03 PM
There are specialized devices that can "print" punch cards
 
I don't own one (also punch cards have lower data storage than paper)
 
@Ginger It would be quite a lot of effort, I'd like to see it in action
True
How big is a "page" of RAM?
 
depends
quick someone find that guy that made a paper on bad swap ideas
 
Like killobytes, megabytes, gigabytes?
So 1 MB
 
wait no one second
according to Wikipedia the smallest possible page size for a 64-bit processor is 4KiB
 
5:05 PM
Oh that's reasonable
You could fit that base64 encoded in a smallish font on A4
 
that's 4096 bytes, or 32768 bits
@mousetail no, this idea would use a matrix-barcode-type-thing for easy readability
 
That would make it even easier
You could even use color for extra depth
 
okay, a Data Matrix code can store 1556 bits of data, so one page would only require a miniscule 22 of them! :p
each code is 144px square
 
hnfuygedhfncsueitghsfneuij
 
uh
 
5:10 PM
@HannahWilliams Hi
 
Hi???
 
sup dude
 
Welcome to TNB!
 
thanks
 
@HannahWilliams Is this some new golfing language?
 
5:11 PM
If you haven't already, please read the chatiquette.
 
Hmmm someone should make one where that prints "Hello, World!"
 
heehee
 
anyway: how many pixels can an A4 sheet of paper hold?
 
5:12 PM
2480 x 3508 at 300 ppi
hmmm
 
Quite a lot in theory but scanners are typically terrible
 
yeah lol
 
You probably want a lot of redundancy so if the scanner just overexposes one corner for no reason you can still read what it says
 
so what's the best way to fit 22 144x144 squares onto an A4 sheet of paper?
they can store 1556 bits, but we'll store 1024 of data per code and have the rest be metadata
 
5:13 PM
cats
 
uhhhhhh
yes???
 
Of course, place the bocks in the shape of a cat and a dog
 
Are you okay???
or just spamming
because we don't allow spam here
 
i am scared
 
5:15 PM
WHAT ARE YOU AFRAID OF??? ACCORDING TO [[Encyclopedia of]] [[Being Afraid]] THERE'S NOTHING TO FEAR EXCEPT
 
i am about to sit next to my crush at lunch and he knows I like him
 
...
Please discuss this in Off-Topic TNB
 
Weren't we gonna make OTTNB official?
 
eh
brb
 
5:42 PM
@TheThonnu When dzaima asked, it produced slightly incorrect code which I fixed and then golfed
A frightening experiment of AI's reliability
Also we had an argument in the Discord which we kinda fixed
 
@Ginger This is just swap space but worse
 
@user ...which is why I called it "Paper Swap" :p
 
Although I guess in this context, worse is better :P
Oh
I'm dumb, ignore me
 
What's the context here?
just.... want to know
 
46 mins ago, by Ginger
When RAM fills up the computer starts printing memory on actual paper using a printer, one RAM page per physical sheet, and when it needs a page you have to put that sheet of paper on the scanner to load it back in
 
5:48 PM
I asked a CMQ asking for cursed ways to do garbage collection
@mousetail Could one of the ROs add it to the room description if OTTNB really is official?
 
oh no not this again
 
Or maybe the chatiquette could be updated
@Ginger Not what again?
 
Making OTTNB official
 
Does anyone have write access to the chatiquette repository?
 
I think we've failed in that
 
5:50 PM
we already had a whole discussion about the officiality of OTTNB
 
How so?
 
and it got nowhere
 
Oh
 
Link?
 
5:50 PM
one sec
 
one of the few Ginger failures
nothing personal
 
7
Q: Should we officially recognize Off-Topic TNB?

GingerRecently, TNB conducted an experiment in which we created a new room specifically for off-topic discussion, which resulted in a marked decrease in the amount of off-topic content in TNB. This room has now been around and active for over five months. Should we "officially recognize" it and add it ...

 
yeah that one
 
It seems like most people want it to be official, given the upvotes on mousetail's answer
 
5:51 PM
Although I still don't understand what exactly "officially recognizing" means
 
Putting it in the room description, I guess
 
> Should we "officially recognize" it and add it to the room description for TNB?
 
Oh ok
 
ChatGPT knows how to golf?!
 
@TheThonnu Welcome to the New Age
 
5:58 PM
@TheThonnu We've had 2 ChatGPT posts here, and neither were golfy :P
 
@cairdcoinheringaahing Ooh that will boost up the case for Adam's MSE post
 
Yeah it's not very good at golfing and its attempts at golfing may break the code
 
@UndoneStudios The one?
 
@cairdcoinheringaahing yeah
 
It's definitely quite far from perfect but at least it sort of knows how to
 
5:59 PM
Wish it wasn't yet
Ok peace off
 
just got Mortarboard on CGSE.Meta!
 
6:20 PM
@UndoneStudios and @all Should I add an "Edit:" section listing sites that are being bothered by ChatGPT answers?
 
Wow, how did you get 10k on meta that fast, iirc you reached 5k a month ago.
 
@mathcat what
oh, adam
 
@Adám Given that the answer is basically "all", I don't think it's necessary
I think we were the last "big" site in the network to get ChatGPT posts
 
> "big"
 
We have multiple questions on the HNQ at any given time, and have decent engagement stats compared to a lot of the ~180 sites on the network
Tho, we have been on a downswing lately, we used to be at around 1k visits/day
 
6:32 PM
@cairdcoinheringaahing But maybe having a list that quickly grows to over a hundred would make them reconsider?
 
@UndoneStudios It would mostly be symbolic
 
if /s is for sarcasm and /jk for jokes, how about /se for dispair?
 
But it would come with things like a link in the room description, maybe an official meta announcement, and probably would be officially RO'd by the TNB RO team instead of just de facto RO'd by them
 
@Adám I doubt very much they'll reconsider
 
Hence "/se"
 
6:34 PM
The official policy is "We'll leave it up to the sites", which is a good call IMO
Note that the posts here that were suspected of being ChatGPT weren't deleted for that reason, they were deleted for being low-quality and for not answering the challenges they answered.
 
why dont we ban chatgpt?
 
because then when it becomes sentient we all die
 
thats a case for the opposite :P
 
@Seggan We haven't needed a site policy so far
 
We don't need to IMO. If it can golf well I invite it to do so :p
It's not like other sites where it could be actively misleading
 
6:43 PM
but it cant golf
 
@Seggan not yet...
 
If the code works it works, if it doesn't it's just a bad post regardless of its source
 
If it becomes a problem, then we'll bring it up on meta. Otherwise, continue to flag, downvote and VLQ posts regardless of if they're chatGPT or just normal crap
 
Remember Dennis? Gentlemen, we can rebuild him. :p
 
tbh i wish dennis never left
 
6:44 PM
That said, if you do suspect a post if from ChatGPT, please do say so in your flags so we get an idea of how big of an issue it is or isnt
 
On DIY or Cooking a bad ChatGPT answer could cause someone to electrocute their cat or get botulism. On CGCC it's just another non-working post, no harm.
 
Is GitHub Copilot only trained on GitHub-hosted code or also on code from, say, SO?
 
Apparently ChatGPT thinks Branf and Piet are golfing languages :P
 
Banning ChatGPT on those sites reduces the amount of work; if it's ChatGPT, delete. On our site, it's pretty much an equal amount of work to check if it works or if it's ChatGPT.
 
@Adám i think gh
 
7:23 PM
@RadvylfPrograms The other concern I've seen mentioned elsewhere is that ChatGPT answers should say they're generated by ChatGPT, in order to follow correct attribution practices.
On CGCC, I think there's been a general policy of "if you didn't write it, you should make it community wiki"--I wonder if that applies to ChatGPT?
 
@DLosc That's explicitly not the case
There's a relevant meta somewhere
 
If you didn't create the post/code, you must provide attribution. It's recommended you post as CW, but not required
 
@DLosc Funnily enough I was actually just discussing the question of ChatGPT and attribution with my AP Research teacher (in the context of education)
My position is that GPT and other AI text/art programs should basically count as people in terms of attribution, both for training and for using their output
 
7:51 PM
@Ginger nothing's stopping you to try and beat the existing sed answer :) if you are interested, have a look on sed's tips page for generating decimal numbers
 
@user41805 yeah, I just thought that answer was clever and it kinda worked :p
 
8:17 PM
opinion: codegolf.stackexchange.com/q/256211 should have received a comment by the close voters - the poster's new to ppcg and even posted to the sandbox
 
 
2 hours later…
10:15 PM
@TheThonnu if only it was that easy for more complex problems :p
Sometimes you'll give it an answer, and it'll golf some of it, you say "golf more" and then it ungolfs what it just golfed and golfs something else about it
Like a game of golfing cat and mouse
 
has anyone given chatgpt a jelly answer to golf?
 
No, because that'd be hell to prompt engineer. :p
It's bad enough trying to get it to do an established esolang like brainfuck, and even worse trying to teach it a standard stack language
I can only imagine the circles you'd be going in trying to explain tacit to it :p
 
2010: Computers are way better than humans at chess, but at least we can still win at Go
2030: Computers are way better than humans at Python golf, but at least we can still win at Jelly golf
 
The disrespect
 
holy shit
 
10:24 PM
That's my Jelly Enigma machine
 
ask it what would be valid jelly code
 
@UnrelatedString Okay, it's spouting some nonsense
 
let's go
 
I'll grab a screenshot in a second
Okay, so it isn't entirely wrong, but the second program is complete nonsense that looks more like J?
 
Lyxal almost managed to teach it vyxal a while ago
 
@emanresuA a very small subset at that
 
The "+/" is correct, the string literal... looks correct, are the delimiters right?
 
Yes
But the +/ would go last, you wouldn't use a literal string, and you'd use S instead of +/
 
Yeah lol
 
@cairdcoinheringaahing That's actually correct J code for the given problem. And Jelly is inspired by J, so…
 
10:29 PM
Copilot vaguely understands vyxal
 
And +/1+1$10 outputs 0210 given no arguments
 
@cairdcoinheringaahing Can you try asking it to do this in vyxal?
 
> I'm sorry, but I am not familiar with a programming language called "Vyxal." Could you provide more information about this language, or perhaps you meant to ask about a different programming language?
 
I'd do a screenshot, but that's effort for such short text
 
10:33 PM
Well I mean it doesn't know Vyxal right? if emanresu had to teach it a bunch to get it to work?
I'd imagine not much literature about Vyxal got included in the training dataset
 
With copilot, it picked it up while I was writing unit tests
 
@cairdcoinheringaahing Why do people tend to post screenshots of ChatGPT instead of quoting in text?
 
For the code section ones, it's a bit hard to do
 
@Adám because the copy paste support is really bad
You try and select the conversation but then it goes and includes things like your email and the time of each message
Which aren't visible when talking to it but are there on the page
Meaning you need to take time to edit those out while a screenshot is just easier
@emanresuA only when it has context
@emanresuA of course it will - it has lots of other data in the file to learn from
And it'll only know it in that context too
 
10:49 PM
Good point
 
even then, it doesn't actually "know" Vyxal or Jelly or any other language it purports to know. It simply writes stuff that it thinks should be statistically completed based on what it's been fed. It may look like it "understands" code snippets, but it's simply the most appropriate/probable token generation from the tokens you give it
 
i think the real distinction here is that it doesn't approximate a systematic rules-based understanding of them, just the end results
 

« first day (4357 days earlier)      last day (484 days later) »