The Nineteenth Byte

Bbrk24

00:13

@lyxal status-norepro

lyxal

Huh looks like a dark mode thing

Bbrk24

There is no dark mode without extensions on CGCC, is there? SO has it but

lyxal

That's what I meant

Bbrk24

Huh yeah, if I enable my dark mode extension then it is colored

noodle man

you can get dark mode on CGCC it just looks terrible

the same theme classes on the body exist on all SE sites

as far as i could tell when i tried it

i guess you would need an extension to toggle the class when the page loads

lyxal

00:17

Strange, it isn't coloured on SO with the default dark

Actually, all the pencils are coloured with the extension

Not just the pencil on that page, but also on comments that have been edited, on the blog posts sidebar thing, etc

Sandbox Posts

02:59

0

A: Sandbox for Proposed Challenges

[Dual scoring criterion challenge concept] code-challenge code-golf fastest-code (that's right...three OWC tags on one post :p) Inspired by The Fast and the Golfiest, I've come up with what might be a novel way to do a combined code golf/fastest code challenge. I haven't decided what the actual ...

Rydwolf Programs

Opinions?

Still undecided on how to handle limiting what languages could be used

Maybe I should just make the flagging rules super strict: no flags which you could remove and still write a working program under

Oh wait, libraries could be an issue

Oh and things like -O3 lollll

Yeah I think I'll just lower 10 to 2, which eliminates the super cheaty stuff (like using flags to inject the whole program's contents), and discourage cheaty flag use

Bubbler

I think I got bit stream I/O working in my lambda calculus interpreter

more specifically, it can now solve infinite I/O challenges

Bbrk24

What's that one language that's just C with a bunch of 1-letter builtin macros

Rydwolf Programs

ooh nice

The Thonnu

@RydwolfPrograms (my opinion is obviously influenced by the fact the Thunno 2 can't compete with the current rules) but I don't think you should restrict languages. Most golfing languages are very slow, so what they gain in code-golf, they will massively lose out on in fastest-code.

Bubbler

03:10

now I have to write functions that parse integers and integer lists out of a bit stream D:

Rydwolf Programs

@TheThonnu If I don't restrict languages you can write a C answer that injects a megabyte of perfectly optimized, ungolfed code

The Thonnu

That's a standard loophole, right?

Rydwolf Programs

Standard loopholes are so subjective they basically don't exist, and I'm overriding default site rules here anyway

IMO the whole loopholes/default I/O needs to be reworked from the ground up

The Thonnu

Just say "no flags", then. I'd rather write a Thunno 2 answer with no flags than write no Thunno 2 answer at all.

Rydwolf Programs

Then a ton of languages just can't compete

The Thonnu

03:12

Like what?

Rydwolf Programs

Tons of languages need flags just to tell the compiler what to compile

Bbrk24

Yeah with no -O a lot of LLVM languages will be massively hurt

I don't know if it's possible to compile C/C++ without -I (unless you rely on the implicit include path)

Rydwolf Programs

And you need flags to link libraries in C IIRC

Bbrk24

yes, you do

Rydwolf Programs

I've changed 10 to 2, so as long as a language exists enough that Hello, World and one other challenge have been answered, it's valid for the challenge

Bbrk24

03:14

In the same vein, dotnet build -c Release --project thing.csproj

The Thonnu

Oh wait I didn't see the any

I thought that said all

Thunno 2 can compete then

Yeah that's all good

Sorry for the misunderstanding

Bbrk24

Interesting concept, but what would the actual problem be?

Rydwolf Programs

Oh hang on...if flag combinations count as individual languages, that might be a problem for things like -O3

Bbrk24

well not really

Rydwolf Programs

Since most C answers presumably don't need optimization

Bbrk24

03:16

some code is only valid under gcc -O1 and not gcc -O0 or gcc -O2

Rydwolf Programs

Well yeah but -O3 -l[some library] -l[some other library] might be perfectly valid but that specific combination of flags likely has never appeared once

Maybe I could just go with caird rules

Sep 9, 2020 at 0:10, by caird coinheringaahing

@RedwolfPrograms Tbf, that was my main gripe with a lot of answers/challenges here. A lot of "technically, ... falls under that, is that allowed?" and I wanted to respond with "Just, ... just don't be a dick and follow the rules in the question"

Bubbler

caird absolutely rules

Rydwolf Programs

"Don't be a dick" with some examples of what's too dicklike (cheaty flags, etc.) could probably work

I think the flaglang gang will understand why cheaty flags are banned here

(Tho really I don't understand whatsoever why a language designer would include cheaty flags other than to be edgy)

Bbrk24

Trilangle has two flags which affect behavior

-f means that it stops once stdout is closed, so that e.g. trilangle -f yes.trg | head terminates

Rydwolf Programs

Since on CGCC flags don't actually give you a competitive advantage

Bbrk24

03:21

and -w prints warnings instead of (or in some cases, before) invoking UB

Rydwolf Programs

And on non-CGCC language requirements are draconian enough you can't really use the flags

I get why the earlier langs like Japt did it, since site rules made it actually save bytes (if in a cheaty manner). Maybe I even get why Vyxal did, to stand out and push some buttons and get people talking about flags

But why would you use or add cheaty flags in 2023? Like what are you trying to accomplish

Bbrk24

hang on I can pull up the full* list of flags one sec
*not counting `--help` and `--version`

Rydwolf Programs

Under official rules you're competing against nobody and under unofficial rules people just add a byte in their heads and you gain nothing

Bbrk24

static CONSTINIT_LAMBDA std::tuple<const char*, char, void (*)(flags&) NOEXCEPT_T> FLAGS[] = {
    { "debug", 'd', [](flags& f) NOEXCEPT_T { f.debug = true; } },
    { "expand", 'e', [](flags& f) NOEXCEPT_T { f.expand = true; } },
    { "warnings", 'w', [](flags& f) NOEXCEPT_T { f.warnings = true; } },
    { "pipekill", 'f', [](flags& f) NOEXCEPT_T { f.pipekill = true; } },
    { "hide-nops", 'n', [](flags& f) NOEXCEPT_T { f.hide_nops = true; } },
    { "show-stack", 's', [](flags& f) NOEXCEPT_T { f.show_stack = true; } },

Bubbler

Japt flag predates flag rule overhaul

Bbrk24

03:24

-e and -D don't even run the code

-n is only meaningful in conjunction with -D

and -d and -s are for debugging purposes

Rydwolf Programs

@TheThonnu Could I get your perspective on why Thunno has them?

Ones like J and H seem explicitly designed for metagolfing (in the metagolfscript sense, not the metagolf sense)

Rydwolf Programs

I wish the term "metagolfing" didn't already have a meaning, since it seems perfect for describing a concept I believe captures pretty much everything I dislike about a whole range of things

Basically, golfing a program by manipulating the language or challenge itself, rather than shortening your code

E.g., offloading work to flags, or twisting an I/O format to save bytes

Bbrk24

I don't remember why I made -f the short name for --pipekill

c--

@RydwolfPrograms CGCC would probably benefit from a BDFL weeding out non-chill answers

Rydwolf Programs

The fun part of code golf is trying to make code as short as possible within a defined interface. Using byte count is how we formalize that. Trying to reduce your bytecount by warping that interface or offloading work to outside of it is optimizing for the wrong thing; it's like the fascination with grades or test scores getting in the way of actual learning

2

(To be honest this is a bigger issue with I/O than with flags IMO; it's pretty easy to mentally offset scores for flags, and we're currently in a nice compromise where both sides are happy. But for I/O, I think everyone would benefit from some rules changes, and I mean everyone)

And IMO I/O and flags are actually the same problem from two different sides in many cases. I think we just need to start treating flags like "join output by newlines" as formatting an untouched piece of data, rather than transforming the data

See: three black boxes

Bubbler

03:36

I think the problem is this: Flags is one interface. I/O is infinitely (or unknowingly) many interfaces.

c--

elaborate?

Rydwolf Programs

E.g., when someone first suggested the (x,y)=> to x=>y=> golf to me in a JS answer, I went with it but it just felt dumb

Like it's the exact same answer. Sure I saved a byte, but all I did was change the I/O to make it more cumbersome in order to save a byte in the most boring possible way

Bbrk24

or when people say to use v-> instead of ()-> and take an argument of type Void in Java

Like I'm relatively more okay with _=> in JS, because in JS you can actually ignore the fact that the argument is there when you call it

Rydwolf Programs

(The solution for I/O that I've wanted for a while is ditching the outdated and clunky default I/O rules, and making ones which are a) flexible and b) refined to fit individual languages and situations using suggestions, with a sort of occam's razor principle that I/O should be as close to the intuitive representation as possible. E.g., no taking a list of characters as a single string where characters are separated by \",\" if that saves you a byte)

So like, there'd be a section for a JS function with rules like "you should take a list of strings as an array of strings", whereas for a full C program it might be "a newline separated string"

lyxal

We need collectives for code golf where language experts can help curate io rules :p

Sandbox Posts

04:38

0

A: Sandbox for Proposed Challenges

Draw this fractal matrix The unique-disjointness matrix ( UDISJ(n) ) is a matrix on all pairs of subsets of {1...,n} with entries $$ U_{(A,B)}=\begin{cases} 0, ~ if ~ |A\cap B|=1\\ 1, ~ otherwise \end{cases} $$ Or a bit less mathematical, it is the 2n times 2n matrix with a 0 in all entries where...

The Thonnu

05:37

@RydwolfPrograms honestly, just because Vyxal has them

And they seem to help Vyxal

I mean, if you have the J flag, for example, it can often save both a "close map" and a "join". And if it's 4 bytes with one flag, people will usually think "oh that's 5 bytes in the old system", which makes it seem better than 6 bytes with no flag.

Idk if you get what I mean, I don't know how to explain it

mousetail

Why not just add flags to your score at that point?

The Thonnu

06:12

Because of the consensus on this site

Simd

07:28

Hello!

Adám

Hi.

lyxal

hi there hello :p

mousetail

3

I thought my keyboard was broken because it kept typing an i instead of an l, but it's just a . from the line above

Simd

I have a graph with 1024 nodes where there is a 10 bit array at each node (one node for each possible array) .Two nodes are connected if they have exactly one bit in common. What is the chromatic number of that graph?

mousetail

2, since 10 is even so every time you change 9 bits you must change the parity

Simd

07:32

Cool. Now what about if two nodes are connected if they have 9 bits in common?

mousetail

Same thing, that's just changing one bit

It's much less confusing to say "change x bits" than to say "keep exactly y bits the same"

Simd

Hmm… I don’t think that can be right

mousetail

Your original problem is only hard because you change an even number of bits

That means the changes can cancel eachother out

Simd

Ahmok

Ah ok

So let me redo my question with 2 and 8

One should be much easier than the other

mousetail

Then I wouldn't know really how to calculate that

Simd

07:39

I need to set up a chromatic number solver

mousetail

Pretty sure that's NP-complete

Simd

It should be ok for small problems

@mousetail it is

But integer programming code is fast these days so can do small problems

Scipy has code for it

For those who don’t hate python :)

As does Julia

mousetail

Well let me know if you find the answer

I've been thinking about your puzzle the whole day but am no closer to a solution

Simd

At least it isn’t trivial :)

I feel the chromatic number part is something people in additive combinatorics would know about

I might ask a math.se question once the bounty is over

Some people spend their life thinking about chromatic numbers :)

mousetail

It's super interesting

PlaceReporter99

08:03

@Simd Just rob Carroll and run away

Simd

08:56

@PlaceReporter99 that's one option :)

When I get time I am going to pose a faatest-code challenge for computing the chromatic number for this puzzle

emanresu A

10:40

@RydwolfPrograms Personally I think that sort of borderline I/O has to be handled on a case-by-case basis by the challenge author. For example allowing taking a list of strings joined by commas might make next to no difference for one challenge (or even be necessary for languages with no concept of arrays), but trivialise a different challenge.

New Posts

11:00

0

Q: Generate the Crystal Maze™️ time matrix

Behold! The Crystal Maze™️ Time Matrix! 0 1 2 3 4 5 6 7 8 9 10 1 0 5 10 15 20 25 30 35 40 45 50 2 0 10 20 30 40 50 60 70 80 90 100 3 0 15 30 45 60 75 90 105 120 135 150 4 0 20 40 60 80 100 120 140 160 180 200 5 0 25 50 75 100 125 1...

PlaceReporter99

12:32

Self-replicating class:

class CT(tuple):
    def __call__(self):
        return CT(tuple(x() for x in self))

class o:
    def __new__(self):
        return CT((self,self))

the class o() is the self-replicating class.

mousetail

New doesn't take self as an argument

It takes cls

PlaceReporter99

@mousetail well, it works, so does it matter?

mousetail

You are creating confusion for yourself

cls is not an instance of the class like self is supposed to be

PlaceReporter99

class CT(tuple):
    def __call__(self):
        return CT(tuple(x() for x in self))

class o:
    def __new__(cls):
        return CT((cls,cls))

updated example

Sandbox Posts

12:51

0

A: Sandbox for Proposed Challenges

Compute this fractal matrix This is one of two versions of this challenge, the one with more upvotes on July 26 will be posted, this version focuses on the computational aspect of the challenge, for the version focused on the artistic aspect see here The unique-disjointness matrix ( UDISJ(n) ) i...

mousetail

Better

mousetail

13:49

CMQ If I randomly bash my head against a keyboard, what langauge has the greatest chance of interpreting it as a valid, program? Program must output something and may not be a quine

Adám

@mousetail One of the golfing languages.

Seggan - On Strike

14:08

^

mousetail

If I add the requirement it must be pristine?

Adám

Still.

The Thonnu

From the golflangs I'd say Jelly/Thunno 2, since their codepage is designed for a normal windows keyboard

So your random characters have more chance of being in the codepage

Seggan - On Strike

14:29

nah, id say Vyxal over Jelly, as Jelly has some weird link ordering quirks

mousetail

Those might increase the chance of the program being pristine

PlaceReporter99

14:45

@mousetail A regular text file

mousetail

I said it must not be a quine

PlaceReporter99

@mousetail bashing my head (actually my hands) generates y5

mousetail

Try again with your head

mousetail

14:59

@cairdcoinheringaahing Can you un-HQN the robbers post? I think only the cops post makes sense as a HNQ

PlaceReporter99

@mousetail 6t

mousetail

@PlaceReporter99 Fantastic

PlaceReporter99

@mousetail a space then 6t

mousetail

Even better

PlaceReporter99

z\ddddddddddddgbnm

mousetail

15:03

Not quite as good

PlaceReporter99

yuuuuhn

;ikucfvghyjuio909i8u7y6t5r3e456789o0p;['

Adám

Can we cut the noise?

PlaceReporter99

@Adám do you have scissors?

The Thonnu

15:27

@PlaceReporter99 Not-Thonnu.github.io/…

PlaceReporter99

@TheThonnu WHAT

The Thonnu

@PlaceReporter99 also vyxal.pythonanywhere.com/…

mousetail

@TheThonnu Nice, it did something useful

The Thonnu

@mousetail lol, I'm not actually sure how it got to that output

And I created the language :p

mousetail

I was just gonna ask

Is that Tonnu 2?

The Thonnu

15:37

Yeah

I made my own site for it

mousetail

Nice

The Thonnu

So that I can have an online interpreter running the latest version

Thanks

caird coinheringaahin g

@mousetail Please raise a flag if you need me to do mod stuff, it makes it a lot easier to keep track of everything

PlaceReporter99

Random number generator (enter seed in input)

mousetail

User interface suggestion

(Slightly reduced the font size of the title and make it say the name of the language literally anywhere)

The Thonnu

15:39

Good idea

Not at my laptop right now

I'll do it tmrw

Also I understand why that's the output now

[ starts a list literal (it's right at the end). It tries to evaluate ['] (it autocompletes brackets) but fails, so just pushes the string. That's the top of the stack, so that gets output.

caird coinheringaahin g

Aww, I missed the Sandbox having it's 420th flag raised :(

mousetail

16:00

Sebastian Lague is hosting an atomic-code-golf chess tournament that might interest some of you: youtube.com/watch?v=iScy18pVR58

New Posts

16:11

0

Q: "Print this diamond" gone beautifully wrong

Write a function (or a whole program) that outputs (or prints) the following ASCII art: Output: 1 1 2 1 1 2 3 2 1 1 2 3 4 3 2 1 1 2 3 4 5 4 3 2 1 1 2 3 4 5 6 5 4 ...

code-golf ascii-art kolmogorov-complexity

Ginger

16:23

@mousetail oooh

~~quick, flood it with submissions~~

PlaceReporter99

let me try something

easter eggs

easter eggs

:(

mousetail

@Ginger If they are interesting I'm sure Sebastian would appreciate it

New Posts

16:59

0

Q: Tips for golfing in Thue

Anyone got tips for golfing in Thue. One tip per answer, and no tips aplicable to all languages (like remove comments)

code-golf tips

The Thonnu

17:15

@NewPosts Python beat JS?!

No way!

Bbrk24

Did you just post the same thing in three languages

Bbrk24

17:38

A literal translation into Swift is approximately way too long

String(repeating:count:) is just so verbose. I like it for legibility but not for code golf

caird coinheringaahin g

18:54

A (joking) sneak peak into SE's new mod policy :P

mousetail

Wait till the AI goes on strike

6

DLosc

19:10

@PlaceReporter99 If you want to test formatting of chat messages, please do it in the Sandbox room. :)

@cairdcoinheringaahing I, for one, welcome etc.

Bbrk24

19:49

try:
    import numpy
    NUMPY_AVAILABLE = True
except ImportError:
    NUMPY_AVAILABLE = False

what am I looking at

DLosc

Some code that tries to import numpy :P

Bbrk24

Why does this Python code have JavaScript in multiline string literals

several hundred lines of it

noodle man

how should answers which don't attempt to be submissions be handled? just saw codegolf.stackexchange.com/questions/59670/the-drunken-bishop/…

which flag reason should be used?

Bbrk24

Check the pinned messages :P

2 days ago, by Rydwolf Programs

PSA: Please don't use VLQ/NAA flags on invalid answers. I've been seeing this a ton in the LQP queue recently. If an answer is invalid, mod flag it, as if it ends up in the low quality queue, either nothing will happen or it will get deleted without bring properly handled. If you see an invalid but otherwise okay (i.e., not spam and contains code) answer in LQP, mod flag and vote "Looks OK".

I also like that they're trying to use C++ in the JS section of a snippet

It's also not valid for lack of includes and usings

noodle man

20:34

@Bbrk24 that doesn't really answer my question since it has code but isn't necessarily "invalid" as much as "not a serious submission"

Ginger

21:18

@cairdcoinheringaahing if this is a joke it's not a very funny one

Rydwolf Programs

21:45

~~I'm all for it~~

It would mean lyxal can get the mods to play chess with us :p

3

And we could have Snowpaw as a moderator

Ginger

~~YES~~

Snowy for CEO 2024

Rydwolf Programs

in Snowpaw's Den, 10 secs ago, by Snowpaw

@RydwolfPrograms Oh h-hewwo there, Rydwolf Pwogwams! giggles It seems you've encountered a w-wascally user using a sockpuppet account to set up a voting w-wing! Nya~ that's against our pawlicies, so I must kindly ask them to stop immediately. No-no voting w-wing for cutie pies like us, okey-dokey? OwO

Ginger

in Snowpaw's Den, 24 secs ago, by Ginger

see? excellent behavior

caird coinheringaahin g

23:23

@RydwolfPrograms Oh, so I should nuke this bot, good to know

Transcript for