The Nineteenth Byte

anyway I think that on linux you need at least 140 bytes to do something (just need syscalls after all), and on windows it's more like 1k (as you need to link parts of winapi)

Lembik

@CensoredUsername 1k! Aargh

a symblink is smaller as I suggested :)

CensoredUsername

If you want to read about compression methods, start here: code4k.blogspot.nl/2010/12/…

with fun things like an 18-instruction x86 asm arithmetic coding decompressor

Lembik

that's really quite bad

how do they squeeze linux onto embedded systems?

in the days of the ZX80 it was all so much more efficient ;)

CensoredUsername

I mean, I work with embedded systems that are <10 euro and have 256kB of program memory

flash chips aren't exactly that expensive either

Lembik

12:15 PM

right.. I suppose no one codes for 16k systems any more

CensoredUsername

could always get the cheapest attiny you can find

2kB instruction mem, 128 bytes RAM

Lembik

aha!

then the overhead might still be annoying

CensoredUsername

Used those a few times, and you can still do plenty of stuff with that

Adám

Am I allowed to dream?

CensoredUsername

well if you're going embedded there's no need for linker tables or executable headers

you need an interrupt vector table and code, that's all

Lembik

12:17 PM

oh I see... so really it's not the same as writing an executable that runs in linux

CensoredUsername

course not

why would you run linux on something that doesn't even have the stack space to do a task switch >_>

Lembik

more like writing for the zx80 :)

no I didn't mean you run linux on the tiny system

CensoredUsername

anyway linux is pretty efficient still

2

Lembik

I was trying to understand why a minimal executable in linux can't look like a minimal executable on an embedded system

Downgoat

@Adám depends, if you are dead, no, otherwise sure

CensoredUsername

12:19 PM

there's a few reasons basically

in embedded executables, you compile code for a very specific target. It is usually set in stone at which memory address your code will be

next to it there's no silly stuff like sections in those executables, because memory protection is for people who actually have page tables

and stuff like attiny's are harvard architectures anywya

Downgoat

@dzaima use async/await

CensoredUsername

in a full operating system however, an executable needs a header section to tell the OS what it is

this includes stuff like "magic number, what memory address to put each section at, which protections to use for those sections, what processor mode the executable should execute in (x86/x64), where the entry point is, what kind of file it even is (relocatable, executable, etc)"

Lembik

this is very interesting, thank you

CensoredUsername

and on x64 a bunch of these fields are 64-bit wide so that header will take up some space

Lembik

got you

CensoredUsername

12:24 PM

now on windows there's the extra complication that a linker section is needed to do anything

Lembik

ok going back to the other question.. do you know if you can call curl (say) in assembly in a small number of instructions?

CensoredUsername

on linux, to interact with the OS, you just set some registers and do a syscall

sure. just call the syscall to execute a child process

Lembik

@CensoredUsername how many instructions is that?

CensoredUsername

depends on the exact syscall

at minimum it's 2, being "mov rax, syscallnumber; syscall"

(for x64 that is)

of course you might have to save other regs beforehand, and if you want to do it right you have to handle a bunch of things

like call the syscall again if you get EAGAIN back

(which is basically "kernel is busy atm, try again"

ToxicFrog

From a C perspective system() is easiest (stuff pointer to string into register, call), but that's section 3; if you want to stick to section 2 you need at least two syscalls (fork, exec) unless invoking curl is the last thing you plan on doing, I think.

execve(2), rather

@CensoredUsername yeah, I'm just not sure if the goal here is "minimize user-written instructions" or "minimize total executable size".

CensoredUsername

12:40 PM

true. either way vfork->exec doesn't require that much instructions either

Leaky Nun

CMC: does the given positive integer have more 1 than 0 in its binary representation without leading zeros?

CensoredUsername

lambda a:"0"in bin(a)

Leaky Nun

@CensoredUsername nope.

Adám

@LeakyNun APL: >/+/1 0∘.=2⊥⍣¯1⊢⎕

Leaky Nun

@Adám so lengthy

Erik the Outgolfer

12:45 PM

@LeakyNun Jelly, 6 bytes: BĠL€</

CensoredUsername

misread it. Python: lambda a:bin(a).count("1")>=bin(a).count("0")

Leaky Nun

@CensoredUsername nice >=.

@EriktheOutgolfer what does it output for 15?

CensoredUsername

leading 0 in binary repr so needed to account for that

Adám

@CensoredUsername Shouldn't it be > rather than >?

CensoredUsername

no because bin(5) == "0b101"

Adám

12:46 PM

Ah

Erik the Outgolfer

@LeakyNun 4, which is a truthy value in Jelly

Mayube

@LeakyNun can we be passed the binary representation of the number?

Adám

@LeakyNun APL: >/(⊢∘≢⌸2⊥⍣¯1⊢)

Leaky Nun

@Mayube nope

Mayube

welp, that's braingolf out then

Leaky Nun

12:47 PM

@Adám that's shorter

@Mayube isn't it turing complete?

Mayube

@LeakyNun idk, probably. Not saying it can't convert a base10 int to base2, but fuck it'd be lengthy

Adám

@LeakyNun AGL: >/⊢∘≢⌸∘Ñ

Fatalize

@LeakyNun More or strictly more?

Leaky Nun

@Fatalize strictly

if I have 4 apples and you have 4 apples, do I have more apples than you do?

Fatalize

If you're French, yes

you also have less

Leaky Nun

12:50 PM

@Fatalize what?

Uriel

@Adám AGL?

Leaky Nun

si j'ai 4 pommes et tu as 4 pommes, ai-je plus pommes que toi?

CensoredUsername

Long live turing tarpits for bit fiddling

Adám

@Uriel github.com/abrudz/agl/blob/master/AGL.dyalog

Fatalize

@LeakyNun In French math, "plus que" (more than) means ≤, not <

Erik the Outgolfer

12:52 PM

@LeakyNun CJam, 12 bytes: {2b$e`0f=:<}

Fatalize

< is "strictement plus que"

Leaky Nun

@Fatalize answer my question

@EriktheOutgolfer fork?

Fatalize

@LeakyNun 7 bytes: ḃoḅlᵐ<₁

Erik the Outgolfer

@LeakyNun what fork?

Leaky Nun

@EriktheOutgolfer is it a fork of the Jelly one?

Erik the Outgolfer

12:53 PM

you mean port, well, yes

but this uses rle

Leaky Nun

I see

Erik the Outgolfer

2b -> B, $e`0f= -> ĠL€, :< -> </

Uriel

@LeakyNun DigitCount[#,2,0]<DigitCount[#,2,1]&

oeis.org/A072600

Jim

@LeakyNun Pyth, 13 bytes: >ssM.BQ/l.BQ2 or >/.BQ\1/.BQ\0

Using K could maybe save one or two bytes

Mr. Xcoder

Why do people ignore this great challenge?

5

Q: How much Mana do I need?

Dungeon Master was one of the first ever real-time role-playing games, originally released in 1987 on the Atari ST. Among other exciting things for the time, it offered a rather sophisticated spell system based on runes. Your task today is to write a program or function that evaluates the number...

Read it till the end, and you'll see it's not that ~~hard~~ complicated.

Jim

1:09 PM

What's the nuance between "hard" and "complicated" in this context?

Mr. Xcoder

@Jim "hard" was misused. I meant "Complicated" as in "long, bored to read it all, need a TL;DR"

Step Hen

@Jim Hard means it is hard to figure out and solve, complicated means that it takes a long time to figure out, but isn't too difficult to solve

Uriel

the best challenges can be read and understood in less than half a minute

5

Mr. Xcoder

It really ain't hard.

Jim

@Mr.Xcoder @StepHen Okay, got it

dzaima

1:16 PM

oh wow all the power rune names are in the SOGLs english dictionary

Erik the Outgolfer

@dzaima I think sogl would be better at kolmogorov-complexity challenges though...

dzaima

it is, just trying it for why not

I would've liked a challenge to draw that base cost ASCII table :p

Uriel

does abusing specific interpreter overflows and bugs considered a loophole?

Adám

@LeakyNun Interestingly, that question can be rephrased as: Return the most common bit in the no-leading-zeros binary representation. (1 if equal)

CensoredUsername

meh, challenges that take 30 seconds to understand usually don't require any real thought to figure out, only code reduction. I like more algorithm-focussed challenges more (with answers in the 200-500 bytes range)

4

Uriel: as long as you mention the specific implementation and said flaw has been present since before the challenge I think it's fine

Uriel

1:23 PM

@CensoredUsername not true. understanding the problem doesn't mean a quick solution

Erik the Outgolfer

@CensoredUsername you can use @at_least_3_chars_of_username to ping other people...

CensoredUsername

I'm aware

Mr. Xcoder

And why do you use name: then?

Jim

@CensoredUsername This one takes 30 seconds to understand, and yet isn't trivial

Erik the Outgolfer

username: doesn't ping

CensoredUsername

1:24 PM

cause I dislike being pinged for trivial answers that do not require a quick response myself

Uriel

@CensoredUsername e.g. can every problem whose solution can be verified in polynomial time be solved in polynomial time?

I bet that would be unanswered long after TETRIS

CensoredUsername

I know there's good 30s challenges, I'm just saying that usually they do not require it

Step Hen

@CensoredUsername The community consensus, at least when I posted my thoughts on meta, is that more people like easy to solve, fun to golf challenges more than hard to solve challenges (more people like golfing than solving)

Mithrandir

Over on Puzzling people like hard to solve things, though ;)

Step Hen

@Mithrandir There you go, have at it

Mithrandir

1:36 PM

(I don't code. I'm here because I clicked on a link. Don't mind me.)

Step Hen

@Mithrandir That's not code :) that's Game of Life

Mithrandir

Oh, I recognize Joe. Z.

...but I still have no idea what to do on this site in the slightest (except for if someone wiggles up a Scratch challenge. Might poke my head in to watch then.)

Downgoat

thag moment when project is actually in python when it said JS

Mithrandir

That goat is right side up.

Poke

debatable

HyperNeutrino

1:41 PM

@Mithrandir Hi! Haven't seen you around here before :)

Mithrandir

o/

HyperNeutrino

o/ welcome to TNB :P

Mithrandir

Thanks :P

Maybe I'll get to know our gamers a bit better.

(speaking of which, been a while since there was a Spyfall game - anyone up?)

Step Hen

@HyperNeutrino btw thanks for the picture yesterday, I think it looks decent

HyperNeutrino

@StepHen No problem! Looks great 10/10 :P

Step Hen

1:53 PM

CMC: Close exactly one open window (there will always be at least one), not including what is running your code

HyperNeutrino

Idea: Press Alt-Tab, Press Alt-F4 :D

Step Hen

@HyperNeutrino Humans do not count as code :P

HyperNeutrino

Too lazy to turn it into code :P

Mr. Xcoder

@StepHen open(window(named:'The Nineteenth Byte')) - That way you are sure it is always open

At least on Dennis' computer, it should run perfectly ^

Step Hen

@Mr.Xcoder Pseudo code: Open TNB, run code blocks in console until an error does not occur

HyperNeutrino

1:56 PM

KeyScript, 37 bytes:

SEQUENCE 18 9 9 9
SEQUENCE 18 115 9 9

Mr. Xcoder

Mac: open(Safari(url: 'https://chat.stackexchange.com/rooms/240/the-nineteenth-byte'))

PSEUDOCODE ^

Erik the Outgolfer

@Mr.Xcoder too ungolfed imo

HyperNeutrino

Actually my code isn't guaranteed to close exactly one window.

Erik the Outgolfer

maybe open(Safari(url:'http:chat.stackexchange.com/rooms/240'))

you'll be autoredirected to https

Step Hen

@EriktheOutgolfer You can specify http: but skip the //?

TIL

Mr. Xcoder

2:02 PM

URL shorteners are disallowed in general.

HyperNeutrino

Not a shortener.

Mr. Xcoder

SHORTENER: https://goo.gl/x4T9SQ

For TNB

GTG

CensoredUsername

@HyperNeutrino: nice attempt at my challenge, I see you also went for recursive descent in the end?

HyperNeutrino

cya o/

@CensoredUsername Thanks :) Yes. Ascent/backtracking wouldn't work easily because of the cannons :P

CensoredUsername

yeah

Stewie Griffin

2:04 PM

Comments? <- Sandbox post. Is it post-worthy? Or is the idea just rubbish (four people have commented, but no upvotes, so I'm not sure).

HyperNeutrino

Anyway, I got to go too. o/

CensoredUsername

although I can give you a slight spoiler: I took a hand at golfing my reference code for it, which is only 525 bytes of python 3

I think there's still a lot of improvement possible ;p

Adám

2:20 PM

@StewieGriffin Looks very nice. Couple of things:

I think you should allow the numbers to be whatever is the native format for the answer language. E.g. APL and TI-BASIC use ¯ for negative numbers, and J uses _.

dzaima

2:32 PM

idk what did I do and now my local SOGL interpreter doesn't work

Erik the Outgolfer

@dzaima rollback and forget about it

dzaima

@EriktheOutgolfer idk what to rollback :/ there's nothing I did except add comments and copypaste

ig I could compare what else did I do

Erik the Outgolfer

@dzaima just rollback everything

btw copy-paste might be the culprit, it's too often the culprit

Step Hen

It is theoretically possible to write an interpreter for a turing-complete language in any turing-complete language, correct?

dzaima

fixed it

was testing why my program didn't work, so I broke the interpreter

intentionally

CensoredUsername

2:39 PM

It is theoretically possible to do anything in a turing-complete language, which of course includes interpreting turing-complete languages

Wheat Wizard

@StepHen Yes

@CensoredUsername No there are plenty of things that cannot be done in Turing complete languages, the Halting problem is the most famous.

Step Hen

@CensoredUsername So you can (not should) write a Java compiler in BF?

CensoredUsername

if it has an infinte tape size, then yes

Wheat Wizard

@StepHen Depending on how complex Java compilation is you probably do not even need your language to be TC to write a Java compiler.

CensoredUsername

@WheatWizard ok, to be more specific, you can compute anything that can actually be computed...

Erik the Outgolfer

2:43 PM

@WheatWizard I think it's just complex enough to just not write it and then blame to the fact that your compiler is compiling...

@CensoredUsername so you can compute the outcome of feeding a java program in a java compiler...right?

Step Hen

@EriktheOutgolfer Someone should make an esolang where syntax or runtime errors cause infinite loops, but the program is usually really slow in the first place

ToxicFrog

@EriktheOutgolfer yes (and at that point, what you have is an alternate implementation of the java compiler)

CensoredUsername

that's called java

2

dzaima

sometimes I really hate that I made the online interpreter in processing.js

CensoredUsername

anyway I think it'd be faster to write an x-to-brainfsck compiler, write your java interpreter in x, and then compile said interpreter to brainfsck

Step Hen

2:47 PM

@dzaima What's the regular interpreter in? the same?

dzaima

processing java

Erik the Outgolfer

@CensoredUsername brainfsck is brainfuck which does fsck on your drive?

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Questions for the Sandbox: Is this a dupe? I wouldn't be surprised if it is. Should I add/remove some of the mathematical operands which might make it too easy depending on the list of numbers? Math: Is the result Rad or Bad? code-golfarithmeticintegernumber Input You'll receive two inputs...