The Nineteenth Byte

lyxal

00:00

^

ATaco

head or first I like best

lyxal

utc midnight!

ATaco

first in particular as it works similarly with lists.

Rydwolf Programs

Rust calls it "take" for iterators, which makes sense in that context. "First" or "slice to" is what I'd personally go with

"Head" is often used for the very first item I think, so that could cause confusion

lyxal

I'll go with 90% of those suggestions

Rydwolf Programs

00:01

Ooh are you mapping multiple names to each operator?

lyxal

I am

Rydwolf Programs

That's what I've been planning to do

Altho actually I think what I was gonna do with Ash, and what I like best, is just ASCII-only escapes for non-ASCII operators and no literate mode

Like \.f for ḟ

ATaco

I did Alias Names in Funky2, found it more confusing than helpful honestly.

@RydwolfPrograms Oooooooh

lyxal

@RydwolfPrograms huh, guess we came up with that idea independently :p

Rydwolf Programs

Shortly after I did I found Husk did that too, so I guess it's a common thing :p

lyxal

00:06

Vyxal 3 has (once they're implemented) what I'm calling sugar trigraphs

which is #.<char> and #,<char>

because there's updot and downdot characters

Rydwolf Programs

Also my worry (and what I'm running into with MIS) is that people'd get lazy and golf in the literate mode :p

lyxal

to me, that's the point of literate mode

why struggle with unicode when you have ascii that automatically turns into unicode

Rydwolf Programs

Well no but like, and then not even convert it to Vyxal or Ash or whatever. Just "Vyxal Literate, 28 bytes" or whatever lol

Guess it's less of a worry with golflangs than esolangs tho

lyxal

@RydwolfPrograms make the interpreter automatically give the short form when using literate

Rydwolf Programs

@lyxal Ah, mine were gonna be a bit more general. In addition to \.f/\_f, you'd have ones like \d for ÷ or \cr for ©, etc.

lyxal

00:10

that just sounds like keyboard shortcuts with extra steps :p

Rydwolf Programs

Yeah but you don't need to install them :p

lyxal

who said you need to install them? :p

Rydwolf Programs

Oh right. Well, ideally you wouldn't need to rely on the online interpreter I guess :p

Like if you're using it on, idk, RTO ;p (well actually no it'll support shortcuts but ignore that)

OOH

What if they automatically got converted? Like keyboard shortcuts!

lyxal

well yes that's also an idea I've had

Rydwolf Programs

If you type \.f in the online interpreter it'd just convert it once you typed the next character

lyxal

00:14

oh I was thinking if you use the shortcuts it prints them as non-shortcuts along with the program

Rydwolf Programs

Shortcuts as keyboard or ASCII ones?

Not sure what you mean

lyxal

ascii

Rydwolf Programs

ah. To me the advantage of the auto-converting approach is that you type the exact same keystrokes in an interpreter that supports language-specific features as one that doesn't

lyxal

So if you use a sugar trigraph, it prints the conversion in stderr or debug

Rydwolf Programs

It'll still run as \.f but an interpreter that's optimized for the language can just turn that into the more readable and golfy version for you

lyxal

00:19

Yeah it still runs it as normal, but also prints what it converts it to

So you can copy paste that as your answer instead of the long form

Rydwolf Programs

Yeah. I guess that's an option too, but I don't really see the advantage TBH (other than a reduction in complexity)

lyxal

User convenience

Rydwolf Programs

Compared to auto-replacing in the text box as soon as you type (but not paste) it

Which is what I'm talking about

I mean I guess it could be annoying if you didn't expect it but I can't see any reasons you'd want to keep the shortcut instead of have it convert

lyxal

I'm talking about for offline behaviour

Rydwolf Programs

Oh offline as in the CLI one?

Yeah that'd make sense

Does anyone use the CLI ones tho? I was under the impression everyone just used the online interpreter

Well I guess you prolly do since you're a Vyxal dev :p

lyxal

00:26

@RydwolfPrograms it's good for testing sometimes and when you want to do things that the online interpreter can't do

Rydwolf Programs

Oh right, didn't think about side effects

Bubbler

Technically TIO/ATO do use the CLI ones

Rydwolf Programs

Well, yeah. I guess those're also situations where the auto-converting wouldn't work

Also, have you done any work on TIB? Were you the one experimenting with Python in the browser or was that emanresu?

The in-browser model's interesting, even if it's not super practical

Bubbler

TIB is long abandoned lol. It was probably emanresu

emanresu A

@RydwolfPrograms Yeah the vyxal interpreter automatically does that

And yeah that was me

The main limit is browser memory. Pyodide uses about 10MB of RAM, and I'm suspicious that it leaks some of that. Something like v86 could run pretty much anything (someone got it running lua) but that's a lot of RAM.

Rydwolf Programs

00:34

10MB doesn't sound like much to me

emanresu A

I suspect in future, the only reason to use a server for online interpreters will be that people's computers don't have to do the heavy lifting.

Pretty much anything can be compiled to wasm nowadays, I just don't have the technical knowledge for most of that. At one point I was trying to compile 05AB1E to JS with elixirscript, but I just don't know elixir.

Rydwolf Programs

The option of having no timeouts seems really promising (I mean, I could do that with RTO, but I'd rather not turn it into a crypto miner hosting service too directly :p)

Bubbler

(has no timeout) (burns your computer, so run at your own risk)

emanresu A

One advantage of DSO's model is that the actual source code can be requested from github & similar, since it's cheap and easy to make requests from a browser. pxeger's said ATO is about 10GB, DSO is about 120KB total (not counting the font, which is about 400kb)

Rydwolf Programs

Oh, taht's true

Not like storage costs much anymore tho right

A $4 droplet gets you 25 GB and $70 on Ebay gets you a couple terabytes of redundant moderate speed storage if you self-host

emanresu A

00:40

True

Surge (the provider I'm using) has a 200MB limit though, and it doesn't have any sort of incremental upload so things do take a while

Rydwolf Programs

RTO's going to be pretty storage intense I'm guessing, which worries me

Since I've only got 800 GB, and 4 open bays in my server (which sounds like plenty of slots but when you take redundancy and stuff into account it's not much)

(part of why it'll be storage intensive is that I'm going to keep every layer version to keep things from breaking over time)

(which for big langs with quick update cycles, that's a lot of space)

lyxal

@RydwolfPrograms timeouts are nice though for if you make an infinite loop by accident and want it to end in a sensible amount of time for testing/debugging purposes

Rydwolf Programs

Well there's always the stop button

If there isn't that's a pretty critical design flaw

lyxal

that's assuming the stop button works :p

Rydwolf Programs

RTO: Scope creep, the web app(TM)

lyxal

00:49

oh no i'm talking about the current vyxal online interpreter :p

Rydwolf Programs

Ugh I'm so used to Chrome OS's Ctrl+Shift+I

@lyxal Oh I know

But every time I discuss something relating tangentially to online interpreters I add a new feature

Like a functioning stop button :p

Bubbler

@emanresuA That comes with its own limitations. Sharing data between JS and Wasm requires COOP/COEP headers, which prevents most 3rd party resources from loading into the page

emanresu A

I have no idea how wasm works. I just use the pyodide API lol

Rydwolf Programs

@lyxal (also "oh no i" is probably one of the funnier apostrophe'd moments I've seen. Looks like you just saw some horrifying monster from the corner of your eyes and it snatched you away before you could finish typing :p)

user

01:24

@AndrovT Ah, thanks for the links!

New Posts

02:41

0

Q: Draw a Gameboy Tile

Description The Gameboy stores tiles as 2 bit-per-pixel 8x8 images, thus 16 bytes. Every two bytes is a complete row with all of the High-bits of each pixel in the first byte, and all of the Low-bits of each pixel in the second. Input Input will be exactly 16 bytes, received through Standard IO ...

Rydwolf Programs

03:18

Wait is it not true that ln(xy) = ln(x) + ln(y) once complex numbers are a thing?

That's so disappointing

emanresu A

04:09

Complex numbers are weird

ATaco

07:21

@DLosc Ask and ye Shall Receive

pxeger

07:58

@RydwolfPrograms it's true "modulo" 2π

and Re(ln(xy)) = Re(ln(x)) + Re(ln(y)) is always true though

l4m2

08:22

@pxeger You mean 2pi*i?

pxeger

well the modulo only needs to apply to the imaginary components

l4m2

Also ln(x) has multiple values, and they've same set

pxeger

to me ln(x) returns the principal argument

but yeah

mousetail

@NewPosts I'd like to see someone do this in SCSS or Less

With lots of linear gradients

Neil

10:14

@RydwolfPrograms xy = exp(ln(x) + ln(y)) is also still true

Sandbox Posts

14:26

0

A: Sandbox for Proposed Challenges

Fibonacci Binary Rectangles I was playing with the Fibonacci sequence in binary like so (note that the binary representations are written here from smallest bit to largest bit): 0 0 1 1 1 1 01 2 11 ...

Rydwolf Programs

15:07

CMQ: What (printable) characters would you include/exclude if you were to create your own version of ASCII?

I'd definitely exclude ~, `, @, ^

mousetail

I'd exclude upper case letters

Rydwolf Programs

I'd definitely include °

mousetail

Endless source of confusion

Ginger

I think ASCII is good the way it is

Rydwolf Programs

I'd probably throw in ¤

mousetail

15:09

Add € for sure

Rydwolf Programs

Maybe one of ©, ®, or ™

Probably the latter for sarcasm reasons

Ginger

definitely 🍪

it's very important

Rydwolf Programs

@mousetail Yeah, then you get into "how many currencies should we include"

mousetail

Remove ", also causes endless confusion. Just type ''

@RydwolfPrograms Then remove $

Rydwolf Programs

I probably would :p

I'd remove # too

mousetail

15:11

Ok " and '' look exactly the same, see the confusion

Ginger

nope

mousetail

Remove \

Rydwolf Programs

\ is nice

It lets you do spinning animations with text

mousetail

Then remove /

only one of the 2 should exist

Rydwolf Programs

Altho I guess it does cause unnecessary confusion about which slash is for what

@mousetail Nobody uses \ as a slash IRL tho right? Why would you ever remove /?

mousetail

15:13

I don't really ever use either IRL

Rydwolf Programs

IMO the en-dash should also be included

And maybe the plus-or-minus

mousetail

en-dash just creates more confusion

Looks identical to - in monospace fonts

plus-or-minus is nice though

Rydwolf Programs

They're different enough in Roboto Mono to be differentiable

mousetail

I'd not call that differentiable

Ginger

@RydwolfPrograms what

all of these are useful

the only one I might remove is $

mousetail

15:15

Only because they are in ASCII

Rydwolf Programs

Only because we've gone back and given them new meanings

^^

~, `, and ^ don't even work for their intended purposes

And @ is just some weird shorthand thing that the ASCII people assumed would be replaced with Á in Europe

mousetail

# and @ are really the worst offenders, only ever used because keyboards have them

Nobody can even agree what # is called

Rydwolf Programs

Make an interrobang one of the characters to make Adám happy :p

mousetail

Imagine if interrobang's replaced # as tag markers on social media

Rydwolf Programs

Alternatively, I'd be more okay with ~, ^, and ` if they functioned like combining diacritics

But then we'd also need a combining acute

mousetail

15:20

ASCII really shouldn't have combining diacritics though, it should be as easy to render as possible

Rydwolf Programs

That's fair I guess

Honestly I do think having one or two dummy characters that're meaningless on their own is nice

Maybe @'s worth keeping

Or maybe something less opinionated to replace # as a tag

Ginger

I like @ and ^, ` can probably go

mousetail

Goals for ASCII:
1. Each character must be easily differentiable. So no I and l
2. As easy to draw as possible. Simple shapes, constant width.
3. Able to approximately express words in a solid chunk of human languages, even if not exactly in the way people would normally write it.

Rydwolf Programs

Whose goals are those? Yours or the designers?

mousetail

Mine I guess

Ginger

15:22

two of those three can be done with a well-made font

Rydwolf Programs

Well by not having combining diacritics you're throwing away #3 entirely, and it'd never be possible unless we really committed and had like, two sticks of diacritics and non-latin characters

mousetail

People can still recognize words without diacritics

My goal is not to write the language like people would write it but just make it recognizable

Ideally it would be something more like IPA than english letters

Rydwolf Programs

Also, a "solid chunk of human languages" will never be attainable without a really Eurocentric view of "solid chunk"

mousetail

No English is actually good since it's phonetic. You can approximate many languages in different scripts with it. If ASCII had been chineese that wouldn't be possible

There are quite a few sounds it can't represent though so it's not perfect

Rydwolf Programs

IMO if we could start from the beginning, instead of doing Unicode how we do it now, it'd be better to make it a bunch of charsets which are optimized for particular languages, which could be shifted between

mousetail

15:26

I think that might complicate parsing too much. Requires keeping track of state. What if some data is missing or corrupted?

Rydwolf Programs

That way the eurocentrism of having Latin-1 be first, and twice as compact, and immune to encoding issues, is avoided

@mousetail Nowadays that's super rare. We've got checksums on checksums in transmission and our RAM's stable.

Sure it requires state, but I think it's worth the downsides

mousetail

Depends how you transfer data. Like a UDP packet could get lost. Multiple processes could be writing to a file

Rydwolf Programs

You already need some state for parsing UTF-8

Not as long-lasting, but still

mousetail

Yea UTF-8 is already pain for that reason but you can recover from errors losing only 1-2 characters in the process

Rydwolf Programs

@mousetail In those cases you'd just put the shift instruction before each packet/write

mousetail

15:29

Your method could potentially never be recoverable

Rydwolf Programs

Well you'd be able to recover at the next shift instruction

They'd ideally be every paragraph or so

mousetail

Yea you'd need to prediodically add shift instructions again

Rydwolf Programs

And you can bet with fair accuracy that it's going to be the same encoding before that shift instruction you find as after

mousetail

You'd get the same issues as with the BOM

Rydwolf Programs

And we could have utilities like file that auto-detect likely charsets as a backup when you don't have any clue whatsoever which encoding it is (which would be pretty unlikely) that would look for common patterns

Ginger

15:31

> Following are our secret evil world domina��nly one copy, so don't delete this file!

Rydwolf Programs

E.g., if most of the bytes are in the 0x60-0x70 range with some 0x20s interspersed, you know it's probably the Latin charset

mousetail

@RydwolfPrograms Yea this already exists and is not very reliable

Rydwolf Programs

@Ginger You'd know with pretty high certainty that that's the Latin charset in the middle

@mousetail Sure, but it would only be an emergency backup sorta thing

Which you would never need today

mousetail

But that is also because in this non-hypothetical world a lot of character encodings are very similar. We could design them in such a way to create more noticeable patterns.

Eg. Make sure the most common letter in every script has a unique byte value

Rydwolf Programs

Also, there's the possibility of somehow interspersing it within the characters themselves (like error-correction-code style), but that would add a lot of complexity

And remember, 90% of the time you know which encoding you're receiving data in anyway

Maybe the shift instructions could have a bit indicating whether it's actually a change or not?

That would allow automatically detecting and probably fixing missing/corrupted bytes that just happen to mess up a shift instruction

mousetail

16:05

@RydwolfPrograms Oh shifts could include both the encoding before and after. That way you could decode text if you just had either end of the segment

Rydwolf Programs

Maybe, but that seems like it'd be a bit wasteful to account for such a rare occurence

mousetail

Maybe, I'd expect the shift to be 2 bytes already, then it would be 3. There could also be a shorthand if the encoding on both sides is the same, then it could be only 1 byte

Rydwolf Programs

Oh, I was thinking 1 byte

Wait no yeah

It'd probably be 2 bytes

mousetail

One advantage is you could read a random 1024 block of a file and decode all of it exactly, without needing to read back and figure out the first shift.

mousetail

16:25

Oh Chess is graduating

l4m2

16:43

https://codegolf.stackexchange.com/questions/107964/output-programming-language-name
Do this have a 0-byte metagolfscript solution?
[tag:quine]

l4m2

16:58

But this ISN'T an error

how to word

New Posts

17:15

0

Q: Find opposite string

Your function must accept one string and return the opposite The opposite string is a string where all characters go in reverse order, all letters are replaced with the opposite mirror letter from the English alphabet and are changed in case, and all digits are replaced with opposite mirror digit...

code-golf string number

Rydwolf Programs

17:55

Interesting...Chrome (on Chrome OS) now has a bettery saver mode

Rydwolf Programs

18:18

Why does the Gmail font make smile emoticons so disfigured

"...he says as he is run over by a car, vertically" is what that emoticon communicates :p

And the nose one is somehow worse

emanresu A

18:40

@l4m2 There's Vyxal h, 0 bytes - h is the flag help menu

pxeger

19:10

@RydwolfPrograms Do you know why this happens sometimes?

Rydwolf Programs

Not sure. I've never seen that before

Do you have it installed twice somehow? There's no way the code to insert that could run more than once (unless the whole script was run twice)

Neil

22:54

@mousetail you might change your mind after tough thorough thought though

or maybe I should have said through instead of after

Newly Featured Posts

23:32

14

Q: Print all pandigital numbers

Given a base as input, output all pan-digital numbers. A number is pan-digital if it includes every digit in that base at least once, possibly multiple times. Every number is considered to contain an infinite number of leading 0s. sequence rules apply. Given a base, you may either: Given an inde...

code-golf number sequence

Transcript for