The first full element list for the experimental branch is out

@hyper-neutrino @AaronMiller @UnrelatedString @Underslash @Razetime @Wasif @Ausername @user @AviFS @pxeger @math ^ (sorry for the mass ping, this is one of the few times I'll be doing such a ping for a while)

This might not be the right place to ask this but I noticed Vyxal has a really nice online interpreter running on pythonanywhere.com. I recently made my own language with the interpreter also written in python and I'm trying to figure out how to make a similar online interpreter. The only problem is that I don't know what I'm doing when it comes to getting a website working - did you guys follow any tutorials to get that working and, if so, is there any chance you could point me to them?

I’m on mobile, but it looks pretty good to me

@DanielH. one moment, I just gotta find the tutorial I followed

digitalocean.com/community/tutorials/…

this is good

(the tutorial)

feel free to use my css/html/javascript templates

and take what you want from my own flask application

@lyxal thank you so much! I'm usually better at google-fu but couldn't find anything relevant earlier lol

@DanielH. also, check out hyper's yuno and branch

I took some code from those two repos for my own server

and then also look at AaronMiller's grok interpreter

and then you might also be interested in this tutorial on setting up automatic updating of your server's code each time you commit

@DanielH. just keep in mind that sometimes even I don't know what I'm doing with pythonanywhere/my site, so I may not know what to do in some cases

also, make sure to check the list of libraries pythonanywhere has by default, so you know what you need to pip install

and then on top of that, before accessing site files within pythonanywhere (e.g. using open(file, mode)), insert the following:

THIS_FOLDER = os.path.dirname(os.path.abspath(__file__))
file = os.path.join(THIS_FOLDER, fileName)

obviously you only need THIS_FOLDER to be assigned once, but you would need to create a new file name for each file you want to open using os.path.join

But that's all once you've gotten the site working lol

@DanielH. ^

Thank you! That's all incredibly helpful

no worries ;)

my current implementation of yuno is in JS so the entire thing runs client-side, so it's not too helpful - unless you're willing to use JS instead of Python

my online interpreter server (branch, flurry, etc) has some stuff that might help but do be warned that it's not actually secure AFAIK

Mmm time to start trying to implement Transformers

what are transformers

Postfix operators

ah

Quicks

Adverbs

That kinda thing

ye

poggers

I'm also refactoring the parser too

Because currently it isn't very well suited to transformers

No major changes are needed though

Because everything is already placed into a list of tokens

ah

yeah managing a stack of functions and then just having transformers pop works quite well from my experience

that's the plan at least lol

:)

(:

Excellent

i take it implementing transformers is going well?

Haven't started lol

lol

Starting off simple with the basic stuff

map is the one i always start off with lol

it's also not too hard to see how to implement it for variadic links, and it works well with infinite sequences

yeah

@lyxal Nice

WOW

so many commands

@hyper-neutrino Good job

but where did the two byte diagraphs go?

They'll come.

as opposed to the three byte digraphs

Three byte digraphs could actually exist...

no.....

Three byte diagraphs no....

twos are enough

Like, the spec allows for them.

oh really?

how does that work lol

group them first

øøm or kk5.

wouldn't those be trigraphs

then execute as single command\

y'all do realize what the "di" in digraph means right

@hyper-neutrino TTrue

@hyper-neutrino Maybe not.

oh sorry

Also, we only actually need one closing character instead of four, and two digraph characters instead of five.

digraph just = 2-byte command :p

or like any group of two characters

if it's 3 it's a trigraph

vyxal has a corpus issue right

@hyper-neutrino yuno is ready?

yes it does

Yeah but it's useless

@lyxal corpus has dead links in it

90% of the more common ones are stuff I use only for ascii art

possibly due to 2x-1 changing usernames

90% of my Vyxal answers are ascii-art

@Wasif its structure is set up so i'm just going through the tedious process of implementing a shit ton of built-ins lol

I'm starting to run outt - page 19/23.

I should be sleeping right now, but... infinigraphs when?

I'm glad we have a new ASCII art enjoyer

@AaronMiller :thonk:

what is 2x-1's current username?

wait so how does it work

ghost

when you see a digraph starter

you just read again and if you see another digraph start, you read again, etc indefinitely?

Also, I just realised English is tacit.

show your work

how English is tacit

@Ausername who??

2x-1 = ghostt

RIP

@hyper-neutrino 69 ¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨¨,

nice

That should be rickroll builtin.

lol

but that dosen't look like builtin

holy heck that's a lot of messages

excellent

@Wasif It's a new constant: kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk69

Can we add kṙ to vyxal as the full text of rickroll?

@Ausername yes

Time to go back in time ~3 hours and go to sleep at a healthy time. o/

it should be built into interpreter

It's funny how TNB is dead and here is thriving.

@Wasif I'm yet to decide on those because I moved some of their functions into the overloads

Good idea

Mmm my computer is almost dead so I won't be able to work on the new parser for a while

Because python is very hard to write on mobile

your parser is already very good

@lyxal not so hard, I have written thousands of python code on mobile

@Wasif yes, but it isn't very well suited to having postfix operators

I'm also doing some code clean-up too

Its been a while I have cloned Vyxal last

need to clone again

I'll be putting the new parser on the experimental branch when it's ready

Ok

893 commits

excellent

I honestly have no idea how I made Jyxal's parser half the size of Vyxal's.

VyParse looks like the parser of a standard language

Anaconda parser is even half size of Jyxal

That's cause Anaconda is simple.

I mean, compare Vyxal to Jyxal.

does anaconda even exist

not so much

I am in a huge fix

what to make it

why is your lexer half golfed ಠ_ಠ

thats how I write code

Snake is like that too

It's probably because it offloads a substantial amount of work to the compiler.

I am thinking should I make it into python or JS

doing it in Js would improve my skills

and online interpreter will be better and faster

@hyper-neutrino probably because mine does more sanity checks

i should have probably specified

Like I make sure that strings are grouped before the actual parsing starts

i was looking at wasif's anaconda lexer.py

I don't do that

@lyxal wdym strings are grouped

I just keep track of if it's in a string, number, compressed integer, etc.

Grouping n-graphs are more necessary

like you divide ab`hello`cd into [a, b, `hello`, c, d] and then parse the tokens?

@hyper-neutrino correct

Not doing so caused too many problems

i don't think that's sanity checks i think that's just standard tokenizing/lexing before parsing :p

not lexing before parsing is painful

It works for me...

I really need to learn how standard languages are made....

granted having your tokens set so you can build a parse tree is mostly relevant for normal languages and you can get away with it if the extent of your parsing is like transformers

But then again it leaves dealing with stuff like vß⁽ßßvka to the compiler as a single token.

I'm starting to regret allowing that.

I think charcoal is not transpiled isn't it

It's python.

I have no idea

I know its python

but it looks like it uses AST and other things that standard languages use

Transpiling is generally faster than interpreting, I think.

yeah maybe

but interpreting gives you more flexibility

@Ausername which of your answers has a double map lambda?

Idk

hold on

None, it appears

some of my answers might have nested lambda maps

but not adjacent ones

Someone's used a double map lambda though

Hmm it was me

That's weird

I think SE search don't work with unicode?

No it does, but when I specified my userid it didn't.

very sus

CMQ: how could I rework the online interpreter to be able to accept multiline strings?

@lyxal when you encounter string delimiter set a string flag, and capture everything in between including newlines and when faced another delimiter close the flag and keep evaluating tokens

But looks like this will not go with your group string strategy

@Wasif No because parsing input = completely unrelated to parsing code.

@Wasif AUsername is right. I'm talking about the input box on the website

oh sorry

I thought vyxal cannot parse multiline strings in code

it can

Try it Online!

nice

So this does not work?

No

because it currently splits on newlines

oh

\n in the input works

Looks like you're going to need to tokenise the input...

no

Try it Online!

it works

that's how you input newlines

lol

nice

but \t and other things don't seem to work

@Wasif ?

Try it Online!

make sure you wrap it in "

I am probably the dumbest person around

no

I wrapped in it ``

Can we fix this

It's extremely counterintuitive.

I'm so used to ` as string delimiters that it feels odd using " sometimes

@Ausername sure

I'll make it so that vyxal literals can be evaluated too

well ` is slightly better looking than "

plus it's unique

unique as in most golfing languages use "

yes

Js uses `

> most golfing languages

even python once had ` as a string delimiter

@lyxal True

@lyxal Wait what

Why was ittt removed then?

@Ausername idk

removed in v3

@lyxal when

it was repr

yeah

repr

not string delim

Oh

ah

uh

lol

XD

got'em

nice

@Ausername ask and ye shall receive

Heh, I forgot how cursed the parser actually is

And I'm remembering why I generally try not to touch it at all

it can't be more cursed than jelly's

# Am I really that lazy? Yes.

@UnrelatedString i can attest to this

so yeah i can't really tell what's going on because there's a lot of sparsely commented code but it's definitely way less cursed than jelly

CMC: Write a Vyxal parser in Vyxal. And no E.

@Ausername this is no CMC

M = massive

this is a full on project

@Ausername your challenge made me realise that my current set operations (eg ⊍) are redundant, as they are already implemented by things like F

So that's a free operation then...

Time to make them multiset lol

Seriously

Yep

Just use dingledoopers sum trick

In both v2 and experimental

Also is there an ascii-only way to remove the first item of a list or string?

Beat jelly by 1 byte

(Old challenges are a pain)

@Ausername \ +1u"i

Oh but for a list, replace + with J

Ok thanks

It works by adding an extra item to the string/list and then taking item[1:-1]

Ok

Because i given a list of numbers acts as a slice

Oh true

Is the answer posted?

Not yet

Because I'd like to upvote pls

Is there a way to reverse a list? (It's possible without, just painful)

(again ascii-only)

Does the list contain sublists?

No, just strings

Try it Online!

wrap then vectorise Reverse

Thanks!

Here

Wow

You poor thing

I don't envy you

Congratulations on winning though lol

Winning?

Absolutely not.

Check the leaderboard snippet

#5.

Mine says #1

Oh well I don't really care

I just wanted to see if it was possible

Heh

@Ausername big brain me adding accidentally useful ascii features

Btw how is vectorising i meant to work?

@Ausername a[I] for a in b

Ok

That is, index each item in the first item popped with b

And of course it doesn't do it at all

sigh

Except it does with swapped arguments

Hmm

Hmm'

Does that mean vectorisation is just generally broken?

Yes it does

Try it Online!

Excellent \s

At least it should be easy to fix

Hopefully

The best comment

@Ausername that's a pretty big assumption ngl

True

dear past me: what the hell did you do to vectorisation?!?

(dear future me: what did you do to vectorisation?)

oh wait... I forgot I made _safe_apply take arguments in reverse

very cool past-lyxal

doge job

oh frick how the frick is triadic vectorisation supposed to have it's arguments?

like I know that it's supposed to do f(item, b, c) for item in a

but I mean in regards to _safe_apply

except for the fact that i didnt implement it as so

The way I'm planning to implement vectorisation is just to do a map, but with the stack it takes input from as the current value, plus the rest of the main stack. Then figure out how many items it popped and put that back on the stack.

I mean, it's kinda too late for you now

I just don't want to have to keep track of the arity of everything.

What are the major differences between the experimental branch and master?

@Ausername heck, it was the laziest way for me to do vectorisiation properly

@pxeger transformers, changed codepage (order and new chars), more array stuff and more useful overloads

nice

@lyxal Sure...

@Ausername emphasis on properly

originally i just turned v<whatever> into ƛ<that same thing>;

That's what I thought I was meant to do for Jyxal

nope

doesn't handle dyads and triads properly

It's much more useful this way anyway

Btw are you ever going to fix element counting in minilambdas?

in e branch, yes

for v2, maybe

:0

I'm so proud of my little parser

being able to do transformers like a big boy

:)

Also, it's smaller now

less lines

:)

hmm

he breaks on structures though

excellent

Hey, give the parser a break, he's just a kid

Idea: A (repeat operation until there's one element left on the stack) commands?

Eg. +. would keep popping and adding the first two elements, until you were left with the sum of the stack

You have Π & ∑ which are like special cases

important to note that this is not the same as wrap + reduce in case said operation takes pops more than 2

or pushes more than 1

or pushes 0 or >=2 elements

I'm considering, for my stack-based lang, replacing most array-based commands with a "execute this code in a new stack frame with this array as its initial stack". A bit like eval, I suppose. (and having the array commands operate on the stack itself)

interesting

@AviFS yes

What does yes mean?

@lyxal ^

It means "yes, that's a good idea"

And that "yes, I'll get around to implementing that"

Idea: Second register?

Or multiple registers, where a number literal tells you which-th register to choose

@AviFS it'll be equivalent to {!1≠|