The Nineteenth Byte - 2020-11-13

« first day (3576 days earlier) ← previous day next day → last day (1270 days later) »

2:13 AM

Find your dog's age: 9 per 468

@cairdcoinheringaahing ^

Any final thoughts on this?

2:29 AM

@Razetime Should we handle negative t1 and/or negative t2?

(negative t2 is somewhat covered in the test cases, but the challenge should explicitly state what to do for such inputs)

caird coinheringaahing

2:42 AM

@Razetime 6, although that may violate the “using previous knowledge to gain an advantage” loophole

@Bubbler ok, added

@Razetime Then you also need a test case where t1<0

3:05 AM

@Bubbler added one

3 hours later…

6:23 AM

8

A: Best of PPCG 2019 – Now Accepting Nominations!

Best mathematical insight On this site we often see answers in languages specifically designed for short code, or designed to be fast. Sometimes, a nice golfing trick or speed-up technique surprises us with its ingenuity, beyond the standard use of that language. And occasionally an answer show...

is there gonna be an awards thread for this?

or can I post one?

6:53 AM

Idk, maybe post a meta asking for one

1 hour later…

8:16 AM

@Bubbler ok sounds good

Did it

8:26 AM

0

Q: Regarding CGCC's Best of 2019

The nominations for CGCC's best of 2019 have already been posted here. If it's alright, I can make another post for voting on the nominations and selecting them(I am assuming that selection is done by voting, from the previous year's thread). So if anyone wants to nominate any other noteworthy qu...

discussion best-of

5 hours later…

1:44 PM

@cairdcoinheringaahing looks like your prayers have been answered

caird coinheringaahing

@Razetime I can tell the future :P

Unrelated String

2:00 PM

I find it funny that I spent an hour incrementally golfing down to what you'd already figured out aside from that you'd forgotten ƒ

caird coinheringaahing

@UnrelatedString TBH I've never really used ƒ, just sort of skipped over it in the docs, which is annoying :/

Unrelated String

Doesn't help anything that it's not on the bookmarklet

caird coinheringaahing

The J_. bit is brilliant though

Unrelated String

thanks

caird coinheringaahing

@UnrelatedString What's the bookmarklet?

Unrelated String

2:02 PM

I would have forgotten about float indices long ago if you hadn't used the .ị trick a couple times to great effect

It's a little JS bookmarklet that lets you click codepage symbols to type them (and also provides keyboard shortcuts)

caird coinheringaahing

@UnrelatedString The .ị trick's been stuck in my head since Erik taught me it during my early JHT sessions :P

Unrelated String

IIRC Adám made it

caird coinheringaahing

@UnrelatedString Oh right, that. I just Copy-Paste from the docs :P

Unrelated String

I forget if there's anything Jelly related that points to it or if I just picked it up when I picked the Brachylog one up, since that one is advertised on the readme

yeah the main thing about the bookmarklet is really that since it happens to have documentation in the tooltips, you can double check things like argument order for stuff you already mostly know lol

@cairdcoinheringaahing also doesn't help anything that there's no cumulative equivalent

caird coinheringaahing

@UnrelatedString Well, there are a lot of open quick characters, it wouldn't be hard to add a pull request to the repo

Ðƒ would probably fit with the general usage of the Ð prefix

Unrelated String

2:10 PM

@cairdcoinheringaahing Oh yeah, you did compile a list of them for that C&R

That too

I had assumed it would be Ðƒ before I found that it doesn't exist

caird coinheringaahing

@UnrelatedString I realised that was completely unnecessary as interpreter.py has this on the 10th line:

# Unused symbols for single-byte atoms/quicks: (quƁƘȤɦɱɲƥʠʂȥḥḳṇẉỵẓėġṅẏ

Unrelated String

...I have seen that several times and could not remember

also it's funny how ( is one of them

I legitimately didn't even realize ) was a standalone quick until I tried using parentheses as one might ordinarily use parentheses

I have a question about whether some content fits the codegolf exchange - me and my friends were discussing creating high-throughput implementations of Fizz-Buzz. We came up with some ideas - we wanted to see what other fast implementations people on the Internet can come up with. Is the codegolf exchange a good place to ask such thing? "Write a high throughput implementation of Fizz-Buzz"

caird coinheringaahing

@UnrelatedString I reckon ( should be like ) but it iterates the previous chain over the right argument, not the left

Unrelated String

@cairdcoinheringaahing That's one idea for sure

caird coinheringaahing

2:13 PM

@OmerTuchfeld You can always write up a Sandbox proposal

Unrelated String

@OmerTuchfeld We have a fastest-code tag, so you could try the Sandbox for more feedback

I couldn't find any content about performance/throughput so it made me doubt whether it fits. Now that you mention there's a fastest-code I'll try my luck in the sandbox. Thanks!

Unrelated String

@cairdcoinheringaahing anyhow yeah Ðƒ is probably the call since I'm not entirely sure I actually see cumulative-reduce-with-starting-value coming up super often outside the context of debugging something written with ordinary ƒ, so allocating a single byte could just be a waste--and it's not like there aren't plenty of aliased quicks anyhow

caird coinheringaahing

@UnrelatedString TBH there's quite a few things I want to change/improve/whatever with Jelly, and I've been considering writing an "unofficial sequel" to Jelly for a while now, especially since Dennis went inactive

Might take a look through the outstanding PRs on the repo and add them to my local version

Unrelated String

@cairdcoinheringaahing Genuinely half of how I ended up with the half of an idea I'm working on now is just that Ḣ and Ṫ mutating their arguments bothers me on a philosophical level

and one of the 10 things I'm torn on at the moment is how much I want to rip off Jelly's syntax

caird coinheringaahing

2:21 PM

@UnrelatedString Well legally we can, it's just a moral issue :P

Unrelated String

:P

caird coinheringaahing

@UnrelatedString That's been a pet peeve of mine for years ಠ_ಠ

Even if there were two commands ŒṪ and ŒḢ that didn't modify, I wouldn't mind, because at least they're only 1 link

Unrelated String

I'm pretty much at the point where I'm going to mostly evaluate chains/trains/whatever I'll call them the same way, except that dyad-nilad pairs actually become monads, and then an assortment of quicks can implement some of the more exotic structures I've considered

caird coinheringaahing

@UnrelatedString So +2 would be treated as a monad?

Unrelated String

yeah

caird coinheringaahing

2:24 PM

That's basically what Jelly already does, but that'd make it more explicit

Unrelated String

and then if it could be ambiguous, prioritize binding right arguments rather than left

yeah

@cairdcoinheringaahing interactions with quicks that count links aside, one way Jelly doesn't quite do that is in monadic chain parsing, where +F "fork"s but +×2 doesn't... and honestly that feels like a good thing, for no particular reason

caird coinheringaahing

@UnrelatedString yeah Jelly's parsing rules are eerily good, but there's always improvement :P

Unrelated String

@cairdcoinheringaahing they're shockingly intuitive for how hard they seem like they should be to remember, and more often than not they work incredibly well

caird coinheringaahing

@UnrelatedString Yeah, a lot of the time (especially with dyadic links/chains), I look at it and go "well yeah, of course that's the right argument, how else would you read that chain?"

Unrelated String

I remember maybe a week ago I tried writing something convoluted in purely tacit APL and that did not go well--and that may be 90% because I don't know APL and made it trains within trains instead of using any semblance of a dfn, but it still speaks to how good Jelly chain rules are that you don't really need anything to not be tacit (the rare ⁸/⁹ notwithstanding)

anyhow one other thing I'm toying with is having some special rules around builtins designated as "constraints", since being the Brachylog user that I am I just have to do that

caird coinheringaahing

2:36 PM

@UnrelatedString Combining Jelly and Brachylog could be a very powerful language

Especially for Jelly programs that have to use #. You could just replace them with a constraint determined by the previous links

Unrelated String

("that" being choice points, and not having special builtins marked as constraints, since everything in Brachylog is one input one output with an optional subscript)

@cairdcoinheringaahing Yeah, I'm thinking that for # sorts of things, the approach might be to take the infinite list of results from something then take a prefix of that

or possibly just have one thing that maps over all the natural number then takes that prefix all in one byte, but general sequence rules seem to tend to also just allow infinite output... come to think of it, though, a "find nth output" 1-byter would be a good idea

The current idea I have of the data model is that everything's built around ragged arrays, and choice is an issue of a "0th axis"

and thinking about that plus reading about other tacit languages has got me spinning my wheels about how to best vectorize functions of lists

caird coinheringaahing

@UnrelatedString Stealing Jelly's ldepth and rdepth for each atom might be a good idea

Unrelated String

@cairdcoinheringaahing I never actually thought to look inside Jelly's innards for that kind of thing, but Jelly's depth stuff has definitely fueled my thinking on this

One of the things I'm sort of looking to be able to do is, say, join inner lists on spaces then join those on newlines, like Jelly's K€Y, but without an explicit map operation--the vectorizer would decide to map the K because it would just... prefer not to apply Y to a flat list, I suppose

one of the crazier things I've thought of is actually statically checking types so I can memoize recursive integer functions, so a holistic vectorization process could come naturally with that

needless to say this kitchen sink of ambition is why the last commit was 18 days ago adding a rudimentary tokenizer for an alternate verbose syntax that I'll use for development purposes until I actually put a code page together

caird coinheringaahing

2:52 PM

@UnrelatedString Language dev gets complicated fast :P

Unrelated String

:P

@cairdcoinheringaahing does not help that I just have to do it in Haskell

caird coinheringaahing

@UnrelatedString Oh man

Well if you know how to do it ¯\_(ツ)_/¯

Unrelated String

my first attempt at making a golflang was actually going to be stack-based and implemented in Prolog

caird coinheringaahing

How do you create an imperative language interpreter in a declarative language?

Unrelated String

but after experiences like this except far less to my benefit, it occurred to me that I don't have any idea how I'd make CLP(FD) behave 100% nicely if I decided to use it at all, and if I don't use it at all I may as well just brew my own backtracking in a language that isn't Prolog

@cairdcoinheringaahing Prolog is honestly really sort of between imperative and functional in a lot of cases

caird coinheringaahing

2:58 PM

@UnrelatedString I've never used it, so the only thing I know about it is that it gets called declarative

Unrelated String

Everything is framed in terms of unifying variables and ...whatever it is you do with goals for sure

It is declarative but the execution strategy can be fairly transparent, and then you also have straight up imperative stuff like write/1

Brachylog's w₄ and family actually just append to a global list in order to perform backtrackable output

...just looking at the commit history reminds me, another thing that's sort of driving my language dev is reducing how many errors are possible

the obvious way to do that in a language with choice points is to just have stuff that would hypothetically error fail instead

caird coinheringaahing

@UnrelatedString I'm currently going through Jelly's atoms and quicks to find those that fail/error under certain conditions

Cause for the dyad-only quicks, why not extend their functionality to monads?

Unrelated String

but there's a weirdly large number of errors you can hit in Brachylog because the underlying Prolog builtins for these things also error instead of failing, like division by zero or asserting that something's length is negative

@cairdcoinheringaahing and extend the two monad-only quicks to dyads

caird coinheringaahing

@UnrelatedString There's three actually {, } and ƙ

Unfortunately, I have no idea how ƙ works, so ¯\_(ツ)_/¯

Unrelated String

ƙ is definitely a surprise

I'd certainly expect it to follow the same logic as quicks like €

caird coinheringaahing

3:10 PM

Yeah, I'd expect this to return the same as this

I think it'd be more useful if it grouped adjacent identical items though

Unrelated String

@cairdcoinheringaahing Both have their uses but that just means that both should be available

...and we do also have Ġ for what ƙ does

brachylog looks really cool

Unrelated String

@Razetime it is

it's probably the most aesthetically pleasing looking language

Unrelated String

@Razetime The codepage's extensive use of superscripts, subscripts, and mathy symbols does wonders... as does the allocation of every capital letter to a variable

3:25 PM

appaprently prolog is really cool

but I'm currently trying out rebol

Unrelated String

Fatalize has definitely made a point of sculpting Brachylog's aesthetic

...at least since he switched it off pure ASCII. Some of those old programs look very J-like (no offense to J)

oh, its used to be pure ascii

Like, BQN and Brachylog probably have my most favourite symbols

Unrelated String

@Razetime yeah, the old version is still on TIO because it's not even remotely compatible with what we have now

like, feels kinda wasteful to have a custom character set and not have it look splendid

Unrelated String

...never mind that the hello world program is the exact same

but yeah

3:29 PM

that's the one gripe I have with stax

Unrelated String

@Razetime I really wish custom code pages caught on earlier so there wouldn't be so much CP437

mans just put in CP347 with like 2 characters modified

Unrelated String

there are box drawing characters in CP437

it jsut hurts

@UnrelatedString Like, ESMin looks amazing

Unrelated String

@Razetime wait really

@Razetime yeah

...you know, what's even worse than CP437 is just that one character that Keg uses

3:30 PM

@UnrelatedString github.com/tomtheisen/stax/blob/master/docs/packed.md

no kidding lol

Unrelated String

④

I just die inside whenever I see ④

lmao

@UnrelatedString It's got this fancy 1.5 byte charset

but I can't figure it out from the code notes

Unrelated String

lol

wht does your thing use?

Unrelated String

not decided yet

it's... going to be a superset of printable ASCII

that's about it

3:32 PM

hmm ok

if it's mostly typable I like it

Unrelated String

there is a wiki page for it

@Razetime yeah one of the things I'm looking to target with how I assign builtins to code points is typability of common things with just a standard American QWERTY keyboard, on top of following some kind of pattern with what kind of character does what kind of thing, and of course being reasonably mnemonic

likee, the useful thing about having the apple keyboard layout

is that you have all the accents and diacritics and stuff with long press

saves a lot of clicking

Unrelated String

@Razetime it really is great

so if it works with mac kb layout it's epic

Unrelated String

of course the verbose syntax I'm going to be stuck with before I get the code page going is going to have support through a command line flag, and the plan is to have verbose tokens directly correspond to encoded tokens, so I might just output the encoded version of a verbose program to stderr whenever a verbose program runs, sorta like whatever Charcoal does

caird coinheringaahing

3:37 PM

@Razetime angry Piet noises

haha piet

if only piet golfing was aesthetic

maybe make a popularity contest with piet art

Unrelated String

@Razetime someone ought to come up with something along the same lines where you're just forced to use large regions of colors, and maybe with a certain amount of curvature

I say

make a 99 bottles of beer program, with art theme

say, christmas

caird coinheringaahing

"make a piet visual quine that uses all colours"

last activity 5 years ago

caird coinheringaahing

3:43 PM

> Last seen Dec 12 '14 at 16:21

> 23,194 REPUTATION

wait, has it already been asked?

caird coinheringaahing

@Razetime What, a Piet visual quine?

yep

say printing a number for each color as a matrix

caird coinheringaahing

Kind of

I dream of making a language challenge where it's just difficult enough to be really interesting like Klein Topololyglot

piet visual quine is just insane though

@cairdcoinheringaahing there was another one where you map your code's byte values to color triplets

4:29 PM

So after peeking around the repo a bunch, I realized there's a hidden page which explains ESMin: bennyboy.tech/ESMin/docs/interpreter3.html

why hidden? nobody knows.

I was looking for a way to find accented characters in a page

and guess who answered that question

5

A: chrome search in page: accented characters

As far as I know, this behavior can't be disabled. You could, however, use a text search extension that behaves differently. Type-ahead-find, for example, only finds exact matches.

Unrelated String

@Razetime is that annotated source code

yep

idk why they thought that was a good idea

Unrelated String

<s>guess I'm writing Perhaps in Literate Haskell now</s>

i wish html worked in chat

Unrelated String

oh yeah isn't SE chat markdown actually a different flavor than SE post markdown

aside from a few obvious things being disabled

caird coinheringaahing

4:38 PM

~~like this~~ ---like this---

Unrelated String

three

~~huh~~

well it makes sense to not mess with --

and there's a sense in which it's more intuitive than Discord's ~~

I think ~~ is more of the standard though

Why does the ESMin Canvas have like

no explanation

Unrelated String

¯_(ツ)_/¯

speaking of Markdown

¯\\_(ツ)_/¯

¯\_(ツ)_/¯

4:53 PM

yay

you got it

graphical output is like, really good in js

and japt doesnt have it

ESMin has like one command explained and nothing else

Unrelated String

@Razetime yeah for a second I thought maybe underscore did italics

@Razetime Doesn't Japt have some kind of syntax for literal JS?

well umm not for the canvas per se

Unrelated String

oh

yeah

you can just eval js code in the middle if you know it real well

aka shaggy

but you still need a lot of mental foritude to walk through that

Unrelated String

asm in c style

4:56 PM

asm with c intermixed

and a few gotos for spice

Unrelated String

👌

bennyboy.tech/ESMin/…

I found a demo

but idk what this means at all

Unrelated String

well it is doing something

are none of the involved commands documented?

looking at the website i posted

no, not really

Unrelated String

~~come to brachylog, we don't even have that many commands~~

4:59 PM

I agree

but graphical output

Unrelated String

yeahhhhhhhhh

caird coinheringaahing

come to jelly, we have dots. Lots and lots of dots

with this new regex find thing I actually might

Unrelated String

one of these days I should just try doing a graphical output challenge in Python or something so I can maybe approach considering writing my own graphical stuff

wait, I've already written Jelly answers right?

caird coinheringaahing

5:01 PM

@Razetime IIRC yes

well then I am already part of dot gang™

Unrelated String

never mind that one of the main things I do with programming outside of golf is writing scripts to mangle videos for comedic effect

Aaa I am so unreasonably mad

so much wasted potential in a language

Unrelated String

ḍȯṫ ġạṇġ

caird coinheringaahing

@Razetime Is Mama Fun Roll still around?

5:02 PM

1.25 bytes per char

@cairdcoinheringaahing not really

but I'm gonna spam his issues just in case

Unrelated String

I actually had to pull up the Brachylog bookmarklet for ġ because Jelly's doesn't have it

caird coinheringaahing

> Last seen Nov 5 at 5:03

That could be worse

@UnrelatedString It's an unimplemented command

Unrelated String

yeah

github is also nov 5

so I'm just hoping

@UnrelatedString group-by in husk

You know what, till then I might as well read up on brachylog

hopefully it's tutorial is less confusing than JEely

Unrelated String

@Razetime it's also group in Brachylog

well

it's called "groups"

and "group" is g

but it's essentially group by identity function

wait no that's a different one

5:07 PM

Ok I found mama fun roll's reddit (1 day ago)

Unrelated String

groups is the split into groups of length one

...that was not meant to be parsed as groups of length one

but yeah back to the point ~~what a crazy coincidence that both of the g dots involve grouping~~

ah there

the funny strikethrough ~~strikes again~~

Unrelated String

@Razetime I remember the wiki page on execution control was kinda confusing when I was new but that's the only real rough point, aside from constraint logic

well, now you contribute to it

wow,

Brachylog is the only language with a video intro

I already like this

Unrelated String

I would say I still don't fully understand ` but really I just never bothered to use it

5:11 PM

Ok this guy has the worst mic nevermind

Unrelated String

there are probably a good few longer answers where I should

I'm gonna make video tuts for golfing langs

Unrelated String

oh yeah there's one that I'm starting to vaguely remember

@Razetime I was going to say something snarky about stack languages but then I realized that they all have way different control flow constructs, I/O, type systems, etc.

the top ones really don't though

Unrelated String

@Razetime I guess you have actually used them

I think I wrote like... one Neim answer once?

5:14 PM

type systems are definitely different

Unrelated String

I probably tried 05AB1E once or twice but I don't think I ever posted in it

but really osabie and stax and keg and whatnot have very similar things

Unrelated String

there is that one time I sorta helped Kevin golf an answer by 20 bytes

@Razetime What even is Keg like these days

I've noticed that old Keg code is completely broken

but I can't really find a lot of documentation on how it's changed

yeah thats why vyxal exists

and vyxal is scarily good somehow

probably because it has more experience behind it

Unrelated String

@Razetime that's what happens when you get used to one thing, and you've done it enough to think "this does so many things wrong and I know exactly how to fix them"

5:18 PM

So somehow the only two golfing languages which have decent documentation and are infix are:

Japt and Pip

Unrelated String

is it APL-style infix or do they have precedence rules

I never actually realized either of those are infix

Pip has a ginormous precedence table

Unrelated String

~~you could consider Jelly infix~~

japt is basically whatever javascript is

problem with golfing in japt is sometimes you just need spaces

@UnrelatedString dloscutoff.github.io/pip/precedence

Unrelated String

one of the things I love about just looking at Japt code is that there are spaces
I don't know what the spaces do
But there are spaces

5:20 PM

So what happens is

ab sometimes parses to a.b()

and sometimes a(b)

and theres even more insane stuff when things get nested

so the spaces are to prevent the intepreter from crying

Unrelated String

@Razetime ah

@Razetime that's actually kinda neat

I remember at one point I was considering doing some kind of crazy precedence type inference combo

> type inference

those words give me PTSD

Unrelated String

@Razetime yeah you'd expect it to be easier to tell when Husk can or can't infer a valid type for a program

There's a bunch of sad lonely programs on my github which failed type inference

lost souls

Unrelated String

F

5:26 PM

@UnrelatedString If my computer was a beast maybe it'd work

but a lot of the times it just runs for an hour

Unrelated String

oh?

@Razetime on TIO it usually takes like ten seconds in my experience, if that long

tio is pretty strong actually

but when it doesnt work it just doesnt

Unrelated String

actually I think inference fails are usually faster than actual working programs since presumably they don't get compiled

or sometimes it just times out and you don't know if it's you or the lazy eval

Unrelated String

yep

reminds me, since Perhaps is in Haskell and I fully intend to use infinite lists, I probably want to keep track of whether or not lists are finite so operations that wouldn't terminate on infinite lists can fail instead

5:31 PM

I agree

anyway

night

a bit late here

Unrelated String

night, good talking to you

2 hours later…

7:49 PM

hi all

Redwolf Programs

Hi!

seems my algo question was more tricky than I expected

New Sandboxed Posts

8:31 PM

1

A: Sandbox for Proposed Challenges

tags: fastest-code High throughput fizz buzz Fizz Buzz is a common challenge given during interviews. The challenge goes something like that: Write a program that prints the numbers from 1 to n. If a number is divisible by 3, write Fizz instead. If a number is divisible by 5, write Buzz instead....

1 hour later…

9:38 PM

@UnrelatedString hey give me a break

It was my first custom codepage and my bright idea to fill it was use circled numbers

Thinking "oh, I'll change those later"

@UnrelatedString I'm not raze time, but I can answer this!

I'm too scared to touch it myself

I mean, it's kinda stable but the tio version really needs updating

Unrelated String

that all makes a lot of sense

caird coinheringaahing

@UnrelatedString I really do not get ƙ

Unrelated String

@cairdcoinheringaahing what the fuck

caird coinheringaahing

My reaction

Unrelated String

also want to move this to JHT since it needs to be antifreezed anyways?

2 hours later…

Redwolf Programs

11:18 PM

I just got a random +1 rep notification, but I didn't gain any rep? Was a post I downvoted deleted on another site?

caird coinheringaahing

Does it say anything here?

Redwolf Programs

No, maybe it was another site?

I don't see any changes for any sites, oddly.

11:40 PM

0

Q: Write a Unicode Unpacker

We all know scoring by characters is ripe for abuse. Let's prove this. The Challenge Output the following: "m8V}G=D@7G1lAI,v08` #(hNb0A8T!g;==SVaG5~ g"jUF!bmRY(3I@na?2S{ fJVzo/GQYU%ybpfUq3aG Yza[jc,WJ$bP^7r};Da} V-!Z+Nk:`/poc}d/X:G\ sWX{dbAUv6,i]%RG$hRp ),bd+?/{U1tU[;<;u.Nk ZFPIOzJ/HimL!nexc,ls H...

code-golf kolmogorov-complexity unicode

« first day (3576 days earlier) ← previous day next day → last day (1270 days later) »

Transcript for

Nov '2013

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