The Nineteenth Byte

thejonymyster

4:00 PM

maybe make only negatives be falsy and not 0 :P

@RadvylfPrograms unless you index at 1

Radvylf Programs

Yeah, but 1-indexing is annoying for other reasons

E.g., modular indexing has some weird edge cases

Seggan

@RadvylfPrograms yeah that was the thing i was doing but then realized 0 also didnt work

Radvylf Programs

One option would be having two different measures of truthiness lol

thejonymyster

i considered that as well lol

Radvylf Programs

Fun fact: An idea that I was planning on adding to Ash but never did was a "truthiness flag". Every piece of data d had a second piece of data hidden inside it, which was used to determine truthiness, among other things. So you could have a -1 that's falsy and a 0 that's truthy, or a 0 that's falsy and a 1 that's truthy.

pxeger

4:07 PM

How would you set that flag?

Radvylf Programs

It was going to be allowed to be any data type, so you could "smuggle" additional data out of a function that would be ignored 90% of the time, but could be recovered compactly if needed later on

thejonymyster

in which case, either A) youd have to have a way to check either one or the other or either for the same byte price probably?
B) youd have to have different situations call for different truthinesses, which could be interesting :?
C) something unforseen by i...

Radvylf Programs

@pxeger Different operatiors would handle it differently

E.g., indexOf would set the truthiness flag based on how many matches there were

thejonymyster

thats totally cool btw

why limit it to just a flag

Radvylf Programs

So if @ swaps the value with its truthiness flag, i@ would count matches

thejonymyster

4:08 PM

why not just give all data a bunch of metadata

Unrelated String

that was one of my earlier ideas for perhaps too incidentally
not for truthiness (obviously) but just having metadata that rides along

Radvylf Programs

@thejonymyster I mean, it basically was that

It wasn't just limited to a bool

thejonymyster

(x+1)[0]

Radvylf Programs

Doing it with more than one flag would be kinda overkill though

Unrelated String

still considering it but realized encoding metadata for specifically under makes zero sense

Seggan

4:10 PM

@UnrelatedString i had a variant of Jelly called Quantum Jelly where all data had one of each type of data in the same container

all operators would operate on both bits of data

sometimes on one container using the 2 bits of data

basically a form of golfy parallelism

Unrelated String

ahah

sounds funky as hell

Seggan

4:24 PM

wh... wha... what...

PyGamer0

lol

ah yes another person interested in hacking is following me

just because i fixed a typo in one termux repository

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Is this a squashed series? code-golf string parsing number decision-problem Given a string of digits, determine whether it is the concatenation of at least two ascending consecutive integers. (in decimal, with no leading zeroes) For example, the string 7891011 is valid, because it's the concatena...

PyGamer0

yeah lol

@Adám why are the ¹²³ misaligned with the other numbers??

Radvylf Programs

Because many fonts render those but not the rest of the superscripts

PyGamer0

can you fix that in APL386?

@RadvylfPrograms wdym

Radvylf Programs

4:36 PM

¹²³ are in a different block than the rest of the superscripts

So many common monospace fonts include ¹²³, since they're pretty early on in Unicode

Whereas the rest of the superscripts are way in the U+2xxx I think

So they fall back to other fonts, and you're lucky if they're even monospace at all

thejonymyster

everything mono when

really though is there not a way to get everything to render monospace no matter the font?

pxeger

some terminals do that, yes, but then you get characters cut in half down the middle

Adám

@PyGamer0 I should be able to fix that, yes. Can you log an issue?

Seggan

4:59 PM

in regards to @Steffan comment to this, why is it so?

pxeger

@Seggan @Steffan > considering that half-byte languages just use 0.5, and this uses less than half of its codepage

dzaima

"half-byte languages" would have 4 bits per character, thus 16 different characters. 96 > 16, thus it'd make sense for it to be higher. n*log_256(96) is the correct thing to do.

pxeger

half-byte languages don't use half the codepage of full-byte languages, they use the square root of it (since there are 256 bytes, that's 1/16th)

a 96-"value" encoding uses less than a 256-"value" encoding (normal bytes), but half-byte languages use a 16-"value" encoding which is less still

half-byte langs multiply by 0.5 because 0.5 = log_256(16)

DLosc

Semi-serious question: can I multiply most of my Pip scores by log_256(96) because Pip generally only uses printable ASCII and newline? :P

Radvylf Programs

If you make an official fractional byte encoding, sure

I find it kind of a boring way to save bytes though :p

pxeger

5:05 PM

If there's an implementation of Pip that has an encoding for that, yes

DLosc

@RadvylfPrograms Agreed, thus why my question is semi-serious

I thought we agreed there didn't have to be an actual implementation, tho?

And how would you actually store log_256(96) bytes?

Seggan

bijective base

Fig can actually read that

Radvylf Programs

But that'd round to a number of bytes

Since byte counts are technically only for within-language competition, it doesn't particularly matter that much. It does make sense that if we're starting to make more semi-competitive ASCII golfing languages that we compare them more fairly though, so even if the official scoring is in bytes, comparing between languages can use whatever we feel like since it's never official in the first place

Wheat Wizard

The question is why would you? It would just make pip byte counts confusing.

Radvylf Programs

@WheatWizard For more fair comparison with SBCS languages, and it's not too confusing, Fig's byte counts use MathJax that makes it super clear what the character count is, and what/why the byte count is scaled by

Wheat Wizard

5:12 PM

@RadvylfPrograms It'd be confusing because then answers after a certain date would be scored with a different method.

Radvylf Programs

Oh yeah in that regard I guess

DLosc

True

Radvylf Programs

IIRC usually when languages change how they're scored, it becomes a v2 of the language

E.g., Brachylog moving to an SBCS

(was that the one that did that?)

DLosc

Yes

I dunno... anytime a question about scoring comes up, I keep coming back to, "Are we comparing scores across different languages or not?" And the answer always seems to be "No, definitely not. But also yes."

6

Radvylf Programs

We officially don't, but everyone does unofficially because half of us are golfing language designers

Seggan

5:18 PM

^

DLosc

It's like a black market economy :P

Radvylf Programs

So we need official policies that allow a broad variety of languages and don't lead to pointless fighting, which we basically have, and unofficial guidelines people use in their heads to compare languages

And I think quite often when we quibble about the second (e.g., flags), we forget it's (necessarily) much more subjective than the first and there's nothing wrong with that

Seggan

~~Flags are wrong.~~

DLosc

Aug 17, 2021 at 16:58, by DLosc

FIGHT FIGHT FIGHT :P

Radvylf Programs

I agree, but since they only affect the second type of byte count comparison, it doesn't matter if people do diagree

(not like I'm one to talk about not arguing about flags lmao)

Seggan

5:21 PM

My stance on flags is semi-conservative: flags that literally just shave a single byte off the code are absolutely cheaty

DLosc

What if it's 2-3 bytes :P

Wheat Wizard

If a flag can shave of 2-3 bytes then your golfing language has a design flaw :P

thejonymyster

@pxeger cant we just squish them instead :P

Radvylf Programs

Not necessarily, you might have a type-II flag like Vyxal's r or M that just happens to save a byte multiple times

DLosc

@WheatWizard I mean, Pip has many design flaws :P

Seggan

5:27 PM

~~like using ASCII instead of SBCS~~

DLosc

Before; after

Seggan

noice

DLosc

Actually, golfed: before, after

Part of the deal with Pip's flags, though, is that they were created back when a flag added 1 byte to the bytecount. So the actual idea was that there was a cost to using flags, it was just cheaper to spend 1 byte on a flag than 2 bytes on Jn (or three on J:n).

Mama Fun Roll

quick q, anyone have a shorter way of writing $[x;x;60] in K?

user

Might want to try the k tree or the APL Orchard

Mama Fun Roll

5:41 PM

sure thing thanks

user

idk k but one could do 60,⍨,⍨x in APL

Mama Fun Roll

yeah I'm just looking for the equivalent of something like x||60 or x or 60

user

Oh I didn't see the $

Seggan

@DLosc we should bring back that rule imo

user

@MamaFunRoll You can abuse the "repeat" thingy

Seggan

5:50 PM

maybe make it the default and allow challenge authors to override it

Radvylf Programs

@Seggan No we shouldn't

We changed it for a reason: lots of languages require flags to be able to run

Seggan

true

Radvylf Programs

And also, costing 1 byte each still allows you to smuggle in additional data for free, but hides that under an illusion of fairness

E.g., if I make a language where the flags are just code concatted to the end of the program, it's fair. But 111 with 2 as its flags and 1 with 112 as its flags result in the same program, and I can take advantage of that to store additional information

Seggan

what

Radvylf Programs

haha you found it in the wild

Wheat Wizard

5:59 PM

my profile pic is technically out of date but spiritually it's up to date.

Radvylf Programs

Doorknob's also no longer a mod

Seggan

@RadvylfPrograms wdym in the wild

Radvylf Programs

As in, not linked there by someone in chat

New Posts

6:17 PM

0

Q: Compress Jelly's code page

In this challenge, you'll print out the SBCS of Jelly, a popular golfing language. Its code page looks like this: _0 _1 _2 _3 _4 _5 _6 _7 _8 _9 _A _B _C _D _E _F 0_ ¡ ¢ £ ¤ ¥ ¦ © ¬ ® µ ½ ¿ € Æ Ç Ð 1_ Ñ × Ø Œ Þ ß æ ç ð ı ȷ ñ ÷ ø œ þ 2_ ! " # $ % & ' ( ) * + , - . / 3_ 0 1 2 3 4 5 ...

Steffan

who made the tnb wiki? it's up to date in that it has lyxal and stuff as room owners

Radvylf Programs

Interesting, that must've been added later on

Since we mostly stopped work on it shortly after it was made in August

user

6:32 PM

Is anyone interested in temporarily reviving it?

Seggan

6:47 PM

ill pay you 3.14159 rep

Radvylf Programs

7:29 PM

7 upvotes in the sandbox is the most I've ever gotten, and yet the challenge is also one of my most poorly received on main. This is why people don't sandbox stuff.

5

The average number of votes on my challenges has gone down since I started sandboxing. There are other factors that contribute to that, of course, but I think it's very overrated.

pxeger

@RadvylfPrograms mine got 10 in the sandbox recently, and then only 15 on main

it must be because of a disconnect between why people upvote in the sandbox, and why they upvote on main

either: the people that upvote in the sandbox do not represent the majority of the voters on main

or: people upvote in the sandbox without answering the challenge or seeing any answers, so they may not reflect how enjoyable the challenge is actually

thejonymyster

or naybe people on main are a bunch of haters :P

i guess thatd be the first one though

Aiden Chow

7:45 PM

lol ya one of my challenge got like 5 upvotes in sandbox, but somehow manage to only get 4 upvotes on main, super strange

Steffan

just noticed I have 6k rep

Aiden Chow

@Steffan dang nice

Radvylf Programs

I think it's pxeger's second suggestion, challenges can look really interesting or unique in the sandbox, but just not be that fun to solve. And a bit of the first one probably plays in too, with users on main looking more for challenges that are fun to solve than ones with an interesting premise.

Steffan

Sometimes I upvote on sandbox, but I forget to upvote when I see it in main

All three of my questions only received like 1 or 2 upvotes in sandbox, but way more on main

What I got the most upvotes for was on SO, lol. it was an answer with I think 120 upvotes or so

Radvylf Programs

Solution: Downvote challenges you like in the sandbox, letting people know they will do well on main

Steffan

7:48 PM

lol

Aiden Chow

0_0

Steffan

actually 190 upvotes lol

I lost track

it's where most of my rep on SO is from

emanresu A

SO rep is stupid

Either you get hundreds of upvotes or you're lucky to get one

Steffan

ikr

Aiden Chow

true

Steffan

7:50 PM

I have another answer there with 45 upvotes

Most of the rest only have 1 or 0

Seggan

22 mins ago, by Radvylf Programs

7 upvotes in the sandbox is the most I've ever gotten, and yet the challenge is also one of my most poorly received on main. This is why people don't sandbox stuff.

this looks so weird on the starboard

18 somebody star this :P

Radvylf Programs

pxeger's already done that :p

in Sandbox, Aug 20, 2021 at 13:35, by pxeger

311 This message has so many stars!

pxeger

ninja'd

Seggan

lol

emanresu A

@pxeger What's funnier is that your message has more actual stars than Seggan's pretends to

pxeger

8:03 PM

@Seggan now that I've starred it too, your ruse no longer works ;)

Steffan

apparently the answer with the most upvotes on SO has 34k lol

pxeger

(I removed it from the starboard though, because it's not actually relevant and interesting enough to be worth having there)

Wheat Wizard

9:17 PM

@RadvylfPrograms Another thing can be that posts to main generally need answers to come in to keep bumping them, while sandbox posts get bumped by edits. So if you have a legitimately interesting challenge which is hard to solve, you are going to struggle a bit on main, but that issue doesn't appear in the sandbox, since being hard to solve has little bearing on how frequently you edit it.

user

@pxeger I think maybe it's also the fact that it's acceptable if a Sandbox post has some weirdness in it, because it's assumed that it'll be fixed later, whereas it's obviously not okay for a post on main to have something wrong with it

emanresu A

@NewPosts Regenerate would be perfect for this but it doesn't support non-ASCII

Radvylf Programs

You could generate the Unicode scalar values but that probably wouldn't work as well lol

emanresu A

9:33 PM

I'm making a patched version of Regenerate

Seggan

10:29 PM

Wait Google is a valid language?

thejonymyster

lol

i feel like it shouldnt be the whole website so much as just the google calculator

because i would argue that hte rest of the search results would have to be reversed as well

Seggan

lol i write hte so much too

thejonymyster

tumblr ruined me man nothings been the same since my mom came home and made hte spaghetti

but uh i mean *negated

Seggan

CMQ: where do yall source your golflang compression dics from?

emanresu A

Jelly

thejonymyster

10:39 PM

The all encompassing dictionary of every word

if i were to do one and actually take it seriously id probably just use the scrabble dictionary or whatever i can find on the google

and just like, make sure it has a bunch of extra words as i see fit

Radvylf Programs

@Seggan Ash's first dict was a list of the top 300k words by frequency, second list was based on CGCC posts

Experimented with using Wikipedia

emanresu A

0

A: Compress Jelly's code page

Modified Regenerate -a --sep=, 229 bytes [¡-¦©¬®µ½¿€ÆÇÐÑ×ØŒÞßæçðıȷñ÷øœþ -~ °¹²³⁴-⁾ƁƇƊƑƓƘⱮƝƤƬƲȤɓƈɗƒɠɦƙɱɲƥʠɼʂƭʋȥ]|̣[AB-DEHIK-OR-WYZ]|[A-IL-PR-TW-Z]̇|̣[abdehiik-or-wyz]|[a-hl-pr-tw-z]̇|[«»‘’“”] Attempt This Online! I thought Regenerate would be the perfect tool for this. Unfortunately, it only supp...

Beat Vyxal :)

user

So I was thinking of a challenge where you create n programs, where the ith program outputs the integers from i*n to (i+1)*n. Scoring would be based on how many program combinations you have where interleaving programs together interleaves their output too (the more combinations you have the better, the more programs you have in those combinations, the better). Any idea if it would be a duplicate?

emanresu A

Can't find a dupe but it sounds really hard

user

Yeah lol, I guess I can put it in the Sandbox and make it more manageable

Maybe the output of the individual programs can be whatever you want

Seggan

10:51 PM

@emanresuA Jelly just beat you

emanresu A

The jelly answer doesn't count

user

Yeah it's literally a builtin lol

Radvylf Programs

It's trivial to the point of esssentially being invalid, which is why I posted it as a CW

user

tbh I'd have suggested putting it in your question itself

emanresu A

^

I only got 10 rep from this but got two upvotes???

Radvylf Programs

10:56 PM

@Seggan Using Wikipedia as a dict source could be a good idea, was considering it for Ash

Seggan

just like scrape half of the wikipedia pages?

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Erverse Hte Ifrst Wto Eltters fo Aech Owrd code-golf Inspired by this accidental misspelling If you could read the title, then you already know what the challenge is about. If not, then I'll tell you. Easy: simply take each word (space delimited), and swap the positions of the first two letters...

Radvylf Programs

11:12 PM

You can download wikipedia

I have a copy

It's 20 GB or so, XML formatted

Seggan

whaaa

des54321

11:35 PM

@RadvylfPrograms damn is all of wikipedia that small? i guess just as plaintext that would be reasonable

thinking about it 20 GB of plaintext would be a hell of a lot of plaintext, that does sound reasonable

Radvylf Programs

With images it's closer to 100 TB

You can download that too but on my internet that'd take approx. two months

Steffan

@SandboxPosts Not sure how actually interestnig this is. We already have swap first and last, so swap first and second isn't that different

Transcript for