The Nineteenth Byte - 2022-05-30 (page 2 of 2)

« first day (4139 days earlier) ← previous day next day → last day (709 days later) »

Unrelated String

8:00 AM

i have a feeling that that overload was made somewhat with challenges like that in mind :P

@UnrelatedString believe it or not it wasn't

I just thought it was useful lol

Never seen that challenge actually

Unrelated String

lmao nice

frankly i can see that

There's already greatest common suffix, so why not join on that

Unrelated String

also the wealth of string overloads vyxal has reminds me, i'm sort of thinking about how something similar could work in a golflang where strings are character lists
because i like character list strings in a conceptual-aesthetic way but one of the worst things about jelly is the only times math builtins aren't useless on strings are when they have bugged interactions with the character type just being python strings

Who wants to make a polyglot?

Unrelated String

8:08 AM

so maybe one option could be to have overloads for lists that contain characters and may also contain integers, and have an operator for forcing that overload even if it is all integers

er, numbers in general

ofc having sensible overloads on the character level is also smart but there are only so many sensible character overloads lmao

This is why I wanted to create Velay

Unrelated String

oh yeah i forgot about that lmao

I honestly think it could've been quite powerful

But I never had the time or motivation

Unrelated String

the issue with my "string"-overload-forcer idea is you then need a sensible overload for that for anything that operates on characters or uniformly operates on lists, but maybe it could be a digraph prefix that just happens to correspond to a forced "string" overload for anything where that's sensible

And what if you have to divide a string into chunks of length N, where N is a input-dependent number that could be 1?

Unrelated String

8:16 AM

it would also not make sense in any conceivable way to apply to anything but an atom so yeah the digraph prefix idea makes sense

IMO the best thing to do is just have a string datatype

Unrelated String

@emanresuA that would be a list builtin :)

with strings, though, you basically never want to vectorise into them

so all operations which vectorise non-list-operations on lists (e.g. +) can be overloaded for strings

Unrelated String

my idea is vectorization "stops" at the level where it detects characters

unless it's something with a character overload but there'll probably be like four of those

some case manipulations and whatnot

The problem however is that most operations that you typically think of as being for strings, like replace or split, are also very useful on lists

Unrelated String

8:23 AM

exactly

and for any of those that aren't dedicated list builtins you want to be able to apply them to arbitrary lists some way or another anyways

though maybe i should actually look at some of the vyxal string overloads to see if there are any good ones that can't sensibly generalize to arbitrary lists

8:42 AM

Have full list and vectorising versions of each lol /hj

Unrelated String

yeah no that's actually a great idea

i was actually thinking of saying some string overloads could just be vectorizing versions of full-list list builtins but in the case of replace and split it's probably smarter to have some kind of depth inference thing

> This is why we now need to store the length together with the pointer, duplicating the amount of information stored by the system.

wait so the system already knows the number of items in an array? (in c/c++)

Really, I'd like a good statically-typed golfing language with something equivalent to typeclasses. I think that would solve some of these problems, because overloads and vectorisation are not really needed, because they're defined by the type system

Now might be a good time to tell you about my secret golfing language idea

Unrelated String

honestly yeah

i do feel like heterogeneous lists are more useful than not for a golflang but it might be possible to cover most of the janky niche usecases with a mix of algebraic kinds of data types

@UnrelatedString I don't completely understand free monads, but Wheat Wizard seems to be putting them to good use for that kind of task

Unrelated String

8:55 AM

good point

c.f. hgl

Unrelated String

i don't understand them at all 😎

@pxeger gist.github.com/pxeger/6e63506fff3999f6ee4c2f12f92fd699

@emanresuA there we go, the curtain is lifted

Unrelated String

most of what i've seen ww use them for is just ragged/mixed/arbitrary depth lists of all the same "atomic" type of values but that's definitely way bigger than the question of mixed "atomic" values

ooh smart

thank you lol

honestly the main reason I gave up on that project (having rewritten it from fresh several times) is because it was such a pain to encode programs in

if you stripped away the variable-entropy / divmod encoding, and just used some variable subset of the 256 bytes, it would be much more usable

Unrelated String

8:59 AM

and adding anything would break a large proportion of existing answers so you'd want to get it right the first time

@pxeger Ooh

@UnrelatedString yeah, my idea was to have something akin to perl's -M flags, where each new "version" was added as a separate module

Unrelated String

mmm

@pxeger Wow that's clever, and much simpler than Radvylf's thing

simpler

I'd definitely call it "fourth-generation"

is there a CS SE room? i cant find it

Unrelated String

9:01 AM

@pxeger finally actually read that right and that's not a bad idea

@emanresuA it's not really that simple, and it's a nightmare to write in

I think it doesn't go as far as Radvylf's idea to eleminate redundancy and I'm not sure how nilads will be handled, but cool!

Unrelated String

it's still simpler :P

yeah lol

i think compressing a program into a picture makes 3 bytes into one pixel and i think that is sort of... unfair sometimes (even tho piet for example doesn't use all colors

Do you think we should change the scoring rubric?

9:03 AM

Ofc not

ok, here's some entirely unfinished code: github.com/pxeger/newspeak

Unrelated String

love the name

Yeah lol

It is gonna be a bit of a nightmare for any sorta recursive iteration

How does husk handle that?

Unrelated String

with first order functions

@emanresuA I did have some ideas for that: all literals, digraphs, and grouped functions (which are mapped over or something), are just encoded as integer literals

then you only need one syntactic concept to encompass all of those

9:06 AM

@emanresuA why

and you pretty much can't have invalid syntax (due to unclosed strings, unfilled groups, unpaired digraphs, etc)

I also thought about trying to write an "optimising compiler", but in reverse, so that you can't write things that are suboptimal like 1+1 (you'd have to write 2 instead)

Unrelated String

i've been trying to think about something like that too

Was thinking about writing one of those for Vyxal

But it turns out it's way too complex a project to write on my own, especially in secret lol

It would pretty much substitute known patterns for shorter phrases

Unrelated String

9:09 AM

yeah lmao

@emanresuA this would be syntactically forbidding something that could be automatically golfed

I guess

@pxeger This community has shown time and time again that open-source is the way to go.

Yeah...

I just liked the idea of coming up out of nowhere and cutting jelly in half

But I don't think that was ever going to happen

Unrelated String

lol

understandable

I think existing golflangs are quite close to optimal

Unrelated String

oh hell no lmao

9:12 AM

You could probably scrape 10-15% off existing golflangs

Unrelated String

yeah

except something like this probably could do more in a systematic sort of way

Even Jelly has a vast amount of "programspace" wasted

It's interesting how golflangs have evolved.

Unrelated String

TRULY vast

First gen: hey ho we're shorter than other langs

9:13 AM

e.g., any program beginning with $

Second gen: Bruh, you don't need multichar builtin names

Third gen: Why stick to ASCII

Unrelated String

you could absolutely beat jelly by some largeish percentage systematically by going half as far as this idea on cutting down on waste

any program containing an unparseable nilad is pretty much useless

Fourth gen: REDUNDANCY MUST DIE

Unrelated String

unparseable nilads can be useful but the vast majority of possible programs containing them are totally pointless

9:15 AM

^

I've seen people use them cleverly

Unrelated String

i do feel like it's only ever one at a time though

maybe i've done two once

@UnrelatedString probably what flax will be?

Unrelated String

making an allowance for useful unparseable nilads would probably waste a lot of entropy in a hypothetical efficient encoding though

@Razetime perhaps is sort of that but i'm talking more about if we did something like the idea pxeger just declassified

where you have exactly the functionality jelly has but syntactically forbid anything that doesn't make sense

@emanresuA fourth gen might have a split between "compressed" langs and ones that are more just refinements of third gen style

9:33 AM

Room for emanresu A and Nobody

→ 35 messages moved to Room for emanresu A and Nobody

9:49 AM

wish python had some functions like flatten, windows, chunks etc..

(builtin not libraries)

Hello!

@PyGamer0 are some of those in numpy?

i guess yeah

(all of them are in more_itertools)

I forgot about more_itertools

I am sad about the downvotes for my challenge

I think it's a good challenge!

And involves a little thought

I know one method that will work

YEAHSAIRHAEIH I DID IT

I MADE SOME SH*T ENCODING FOR ASCII

@pxeger wow, it's almost as if that's one of the key ideologies of my popularity blog post :p

9:59 AM

7/8 bit (i think) encoding for ascii

WOW

it works

@graffe which

-3

Q: Sample integers that sum to one hundred

Given a positive integer n, randomly output n non negative integers that sum to one hundred. n will be at most 200. The output should be present as a list of integers (not sorted). Your random sample should be uniformly sampled from all lists of n non negative integers that sum to one hundred. Yo...

code-golf probability-theory

it works compressing a test file by 12.455%:
https://dzaima.github.io/paste#07VN/T1JhFP7/foort/gh5L2kc/UyNrbAdDKDTbem1xUWIpvCgOvEIQtJSW2ZomE4nZoagWOlKZW1/jjnm7yfpHMDpk1bX6Dzx33Pued5n/ec5@w4nQ3r6fMN9ItdPV6P85IJ8pNoZCowbUqIocBEUIyOiuPh0aAQj2oBLchSyUfmpFWatkjTaZUJkWA4NDYSnYwnWCrO7KJiM5ulpOWW3eKoX5CZzIbiw656NBR3JIfTQmJygqWssskqJ9OCzs5SM7LZ7khajO1ih1NlBHBdULsINRnRwuMs9VhhBjWiRgyyQf@YDKJkcOkMKmEEWQsmNCE0Hn4ajDOzIlI9DkWk02En325ppjRb3bFpzGaU6n4zp9hMjaRJqb/qagAE51XlRK@n7w/1bt@xX@A8ffceuD3uqxpLvLALr3gxq2AVFqEieO/iMc76YUXQ4BvfmMNMP@wP8tz6GJbacBOyrbiCK1DDeZiHL2kCZ6Am4UtYa73Jc3k8a4Ut@O7DPcUNLwSoET/Bj/GA…

@Nobody thank you if you upvoted!

@PyGamer0 i guess yall can use this compression for golfscript or k or something lol

10:13 AM

TIL that the entropy of English is ~1 bit/cahr

10:49 AM

CMQ: Will 4th gen golfing languages be strictly fractional bytes or will there still be a place for SBCS?

@lyxal what is

> SBCS?

Single byte character set

Basically what allows languages like vyxal, jelly and 05ab1e to have all their funny Unicode characters as 1 byte per character

SBCS, or Single Byte Character Set, is used to refer to character encodings that use exactly one byte for each graphic character. An SBCS can accommodate a maximum of 256 symbols, and is useful for scripts that do not have many symbols or accented letters such as the Latin, Greek and Cyrillic scripts used mainly for European languages. Examples of SBCS encodings include ISO/IEC 646, the various ISO 8859 encodings, and the various Microsoft/IBM code pages.The term SBCS is commonly contrasted against the terms DBCS (double-byte character set) and TBCS (triple-byte character set), as well as MBCS...

htreya

ok

@Ginger thabks for the engrish very cool kayne

11:03 AM

I needed a new greeting

11:18 AM

chat.stackexchange.com/rooms/136691/a-mathematical-problem

@lyxal so you can make your own set?

pretty much

@lyxal and letters in that set would be 1 byte?

yes

@lyxal can i have say "frick" in my SBCS?

I don't think so about that

11:29 AM

You'd need a rather complex encoding for that

Ooh, now I have a random idea: SBCS but words

2

YOu end up with a program

frick that thing it never is
gotta go the banker reads
pity this is never it

ahh, poem

@emanresuA .

@lyxal why

@emanresuA not a bad idea...

@PyGamer0 haha

ten range sum increment

@Nobody how is that funny?

idk do you have to have a why for humor

11:38 AM

c h e e s e

wait what's that

@PyGamer0 .

Mar 29 at 20:55, by Radvylf Programs

user image

oh no

user image

that means bad conncetion

@Nobody how

12:00 PM

@Nobody You tend to post a lot of noisy messages like this which don't add anything to the chat. Please read through our chat guidelines (especially the section "don't make noise").

CMP: for inline comments: #{ }#, or #{ #}, or #{ }?

12:17 PM

1st if } is otherwise used in the language. 3rd otherwise.

12:39 PM

First , second is kinda odd

third

probably first

third

1:46 PM

Whenever I see a cat in a gen z meme compilation I instantly think of @Ginger

can't believe my broken humour has been ~~tainted~~ infiltrated by TNB

Unrelated coding/physics ebook recommendation

oh that

nature of code

i used to watch the coding train

I think I watched his Java tutorials

i just watched his coding challenges

oh that mac.. we don't talk about that mac, not after the incident.

...

1:55 PM

@PyGamer0 mmhm that's right

Apr 4 at 13:44, by mathcat

what have I done

btw i use ubunz

and i use winduz

Apr 4 at 14:11, by Radvylf Programs

Thank you for making me laugh in class as we discuss medical experimentation in the holocaust

2:26 PM

val range = (start, end) ->
  start >= end ?
  [] :
  [start] + range(start + 1, end);

CMQ: How do I make ^ function tail recursive?

are you asking in the abstract sense, like how a compiler would do it?

no

how do i make the function tail recursive

so the compiler can optimize it?

what language? or is it pseudocode

that is pseudocode

@PyGamer0 Not sure what this language is but maybe something like

2:36 PM

(looks like JS)

(i will implement it if yall want :P)

val range = (start, end, acc = []) => start >= end ? acc : range(start + 1, end, acc + [start])

ok cool

val _range = (start, end, accum) ->
  start >= end ?
  accum :
  _range(start + 1, end, accum + [start]);
val range = (start, end) -> _range(start, end, []);

i will call the pseudocode "Wellscripted" :P

Wellscripted (WS) will transpile to JS

Wellscripted will have the following features:

9 hours ago, by PyGamer0

language be like:
- macros like C
- simple like python
- functional like haskell
- object oriented like java
- pattern matching like scala
- typed like Typescript
- fast like k
- array oriented like APL
- popular like javascript

Oh god no

python is not simple

why on earth would you want macros like C

because yes

2:47 PM

Literally the only thing on that list that would make sense is "functional like haskell" (and maybe "fast like k", I don't have experience with that)

"functional like haskell" but "typed like typescript" => "functional to the extent that JS is functional"

~~javascript is not popular~~

or just not make WS transpiled

just make it in C

WS will be compiled :p

oh and btw WS will not have loops :p

@user pretty sure k is faster than C++ (may be wrong)

@PyGamer0 you can write both slow and fast code in both languages

yeah

2:50 PM

you can write both slow and fast code in any language

different langauges will have different upper bounds on performance, but it's pretty stupid to say "k is faster than C++"

I mean, when someone talks about speed, they're talking about how an average idiomatic k program would compare to its equivalent in C

especially when k and C++ are so different

Also depends on what domain they're used in ig

yeah, writing a control system for an aeroplane in k would probably not work as well as it would in C++

You can probably dip into assembly with both languages if you want to go all out

2:52 PM

Wellscripted:
- fast (probably)
- easy to learn (maybe)
- object oriented
- functional
- array oriented
- compiled (or transpiled)

You might want to rethink object oriented + array oriented

Not that you couldn't make a language putting OOP, FP, and arrays together nicely, but it'd probably be hard

@user what is the problem with that?

I feel like array oriented languages mostly encourage you to keep your data as numbers in ndarrays, while object oriented languages have classes with named fields and stuff instead

in an array oriented language, where do objects fit in the scalar -> vector -> matrix hierarchy?

are they scalars?

i guess a better phrase will be "powerful array manupilation"

2:54 PM

BQN does let you treat your objects as arrays iirc

@pxeger because in that case, arrays are not really first-class objects

@pxeger i guess yeah

how about having no loops for WS

:p

if you're doing a functional language, that's fine

i guess the main focus will be functional / array

loops are a pretty imperative construct

2:58 PM

but the question is: how the fricko do i implement this

start with a lexer

then a parser

then a compiler to a simple bytecode, for now

then a simple bytecode interpreter

ok noted.

Depending on the features WS has, you could even transpile to a high level language until you feel confident enough to write a bytecode interpreter

I think a bytecode interpreter is not too complicated, and transpiling actually can be

a bytecode interpreter is no more complex than implementing a slightly more advanced version of brainfuck

Depends on the language

@pxeger Designing what your bytecode will look like and compiling to it is kinda complex though

3:01 PM

stack-based bytecode is what I meant

I've written a transpiler to Java for a simple, Java-like language before and it was really easy, mainly because the language was already like Java

@pxeger so first i will figure out the grammar for WS

time to pull out the good old pen and paper

and i will also leave

o/

@pxeger cursed idea: Myxal can be the bytecode :P

You will hate yourself if you try that

Compiling to a golflang is a bad idea

that is why its a cursed idea :p

cursed idea 2: a computer in minecraft that runs vyxal

3:24 PM

> flax is a Python library. flax has no bugs, it has no vulnerabilities, it has build file available, it has a Permissive License and it has low support. You can download it from GitHub.

> The latest version of flax is v0.1.4

that is outdated

kandi.openweaver.com/python/PyGamer0/…

> kandi has reviewed flax and discovered the below as its top functions. This is intended to give you an instant insight into flax implemented functionality, and help decide if they suit your requirements.

Tokenise a program into tokens .
Parse tokens .
evaluate a monadic chain
r Evaluate a dyadic chain .
Convert a string to a number .
Scans a list of links .
Fold a list of links .
Filter a list of links .
Calculate a function of a function .
indent a flax expression

indent a flax expression lmao

I wonder what language C is written in...

3:40 PM

C refers to a language standard, so arguably, C is written in LaTeX

curl web.archive.org/web/20181230041359if_/http://www.open-std.org/… | grep -a Creator

/Author(JTC 1/SC 22/WG 14)/Title(ISO/IEC 9899:2017)/Subject(Programming languages - C)/Creator(LaTeX with hyperref package)/Producer(pdfTeX-1.40.18)/Keywords()

what is the time between submitting a question and being able to add a bounty? Is it 72 hours?

@graffe from codegolf.stackexchange.com/help/bounty:

> A bounty can be started on a question two days after the question was asked

is that 48 hours?

I just wanted some precision :)

I presume so

the message "Question eligible for bounty tomorrow" is not at all useful given that we could be in any time zone

3:50 PM

does it give you a timestamp if you hover over it?

unless it means it means "we are currently in the first 24 hour period, after the second 24 hour period you will be able to add a bounty"

you just get "offer some of your bounty for better answers"

if I hover over it

StackExchange uses UTC for (almost?) everything

that's interesting! Are they any countries on UTC currently?

dayspedia.com/time/zones/utc+0 :)

stackexchange is portuguese :)

have we had a challenge to count the number of ways that n non-negative integers can sum to an integer k?

I presume you mean unique non-negative integers?

because otherwise there are infinite

sorry I mean a list of n non negative integers a_1,...,a_n so that sum a_i = k

how many different such lists are there for fixed n and k

3:59 PM

oh

Wellscripted

A theoretical programming language.

yes i made a room

:)

mathematica has a built in for it I assume :)

Is it codegolf.stackexchange.com/q/135875?

That question's description is so barebones that I'm not completely sure, but I think it is

I think so

well found!

I am surprised no one has posted a built in solution to my challenge

or any solution at all!

1 hour later…

5:17 PM

→ 6 messages moved to Add a (hidden) language to a polyglot

6:08 PM

thxz

6:43 PM

@graffe

Is this

valid for your challenge

not sure about the uniform distribution / brute force part

nvm it's definitely not

7:40 PM

@graffe It does seem to be a dupe though

Actuall yno the zeroes

Tfw TIO prints ANSI escapes

@emanresuA the same happens with some languages on ATO

Is it possible to color output instead? (easily, at least)

it would be, with something like xterm.js, but it's not that simple, no

Aww, ok

8:25 PM

0

Q: fastest matrix multiplication on x86

This challenge requires integration with C, so you can stop reading if you're not interested. Matrix multiplication is a simple operation, but the performance depends a lot on how efficiently the code is written. Let's compute m = x * y. The following code is the implementation of the textbook al...

matrix fastest-code optimization c x86-family

Can anyone interpret codegolf.stackexchange.com/a/248004/108721 ?

Radvylf Programs

8:49 PM

Ooh carryless multiplication is really neat

CMC: A000695, the perfect squares with carryless multiplication

New Posts - Challenges - Code Golf: From the smallest seed

posted on May 30, 2022 by Moshi‭

Challenge It's a bootstrapping challenge this time! Write a full program that, once run, writes the source code of another program that in turn, once run, writes the source code for an...

@RadvylfPrograms Vyxal, 7 bytes

Radvylf Programs

9:44 PM

CMC: A014580, the carryless multiplication prime numbers

Radvylf Programs

10:01 PM

I've wrote functions to do carryless multiplication, division, modulo, factorization, exponentiation, and factorial, and it feels like I've been transported to some weird alternate universe where all the math stuff still works but the results are slightly different

Radvylf Programs

10:31 PM

Well today I think I found the worst thing

An infinite scroll dropdown menu for time zones

You see the first 20 or so time zone descriptions, then scroll down and wait a second for more to load, then 20 more, and so on

revenge on timezones for being so plentiful

10:49 PM

@CodidactPosts This could be an interesting challenge, but I think most people will just post trivial quine modifications (me included)

I was gonna post kS in Vyxal

How would that work?

Oh I see

kS -> amogus character -> sus -> ???

kS -> ඞ -> sus -> large error -> sadly ⟨ 1 ⟩

Aww

I'm sure you could modify it to work though

11:43 PM

@CodidactPosts RprogN 2, 6 bytes. «‘ .]

Grows exponentially

@emanresuA why does vyxal have an amogus meme ;-;

@emanresuA ... so what was the bug?

« first day (4139 days earlier) ← previous day next day → last day (709 days later) »

Transcript for

May '2230

The Nineteenth Byte

The Nineteenth Byte: General discussion for codegolf.stackexc...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile