@lyxal because of the edge cases i presented

how should (ak|bc) be parsed?

var = ak code = bc

and how do you plan to do that with your current lexer

yesn't

in the parser

right but

your tokens will be

( a k| b c )

the | won't be a token

ez

lexer ignores | in digraphs

that sounds rather hacky

so presumably you no longer care about allowing non-identifier characters before the |?

I do care

then how about (ab`cd|ef`gh)

what should that do

a, b, cd|ef, g, h

no variable

so no digraphs should support |

and strings should take precedence over |

and non-alnums should still be allowed and simply ignored

correct

hm. letting the lexer just spit out all the tokens and the parser fix the mess still feels wrong but this should work in theory

well (`|`)

should be a string

containing a single |

only outer | should actually branch

Lyxal pushed a commit to Vyxal/fresh-beginnings: Update Lexer.md

i agree; that's just inconsistent with the description you previously had

so what about (ab`cd`ef|ghij)

var = abcdef code = ghij

at least, that's what it currently does

!!/run (ab`cd`ef|ghij) c

[@lyxal: 58742268]
0

STDERR:
global stack, register, printed, output, MAP_START, MAP_OFFSET, _join, _vertical_join, use_encoding, input_level, raw_strings, retain_items, reverse_args, this_function
for VAR_abcdef in VY_range(pop(stack)):
    context_level += 1
    context_values.append(VAR_abcdef)
    stack.append(VY_min(iterable(pop(stack))))
    stack.append(iterable(pop(stack))[0])
    rhs, lhs = pop(stack, 2)
    stack.append(index(lhs, rhs))
    rhs, lhs = pop(stack, 2); stack.append(join_on(lhs, rhs))

w....wait

since when does clicking full text open in a new tab?

is this an actual SE feature or is it because of my userscripts?

oh okay it's just a userscript

Lyxal pushed a commit to Vyxal/fresh-beginnings: Made sure comments and digraphs lex properly

PR #204 (Vyxal/Vyxal) opened by Lyxal from Vyxal:fresh-beginnings into Vyxal:fb-lexer: Fresh beginnings

PR #204 (Vyxal/Vyxal) was merged by Lyxal from Vyxal:fresh-beginnings into Vyxal:fb-lexer: Fresh beginnings

Lyxal pushed a commit to Vyxal/fb-lexer: Fresh beginnings (#204)

→ 12 sussy commit messages that came from an upstream merge moved to ­Trash

Lyxal pushed a commit to Vyxal/fresh-beginnings: fixed a few bits of formatting mucked up by the merging (hello @hyper-neutrino)

hehe double ping

one in the chat and one in the email

hello @Razetime

good morning

you can now lex fizzbuzz

that is the only program you can lex

all other programs are just ignored ;p

genius

@VyxalBot did i do something wrong

@lyxal

@hyper-neutrino no

thonk

I never merged yours kekw

I copy pasted the comment lexing block from the pr

And left a comment on the pr

The merge was me updating the lexer branch from fb

And resolving merge conflicts messed up a few newlines

nice

PR #203 (Vyxal/Vyxal) was closed by hyper-neutrino with unmerged commits.

hyper-neutrino deleted branch Vyxal/hyper/parser-fix

parser fix?

Bruh this is only the lexer

Uh.../

Howdy!

Hi

Is lex even a word?

Yes

yeah i misnamed my branch and couldn't be bothered to fix it

Lol

I'll get started on drafting the parsing specs later today

But in summary (and while I remember this): structures will be dictionaries with the keys being the different branches

When parsing a structure, you concatenate everything to the default branch of the structure until a | is reached

Upon reaching |, you start concatenating to the other branch

The branch that you concatenate to is called the active branch

That should say the starting branch

Because the default branch is for if there isn't any branching

I'll specify it all better later

This is just a high level summary

Ok

@lyxal do you remember what we discussed for how [a|b|c|d|...] might work

(worth considering for the others, maybe?)

@hyper-neutrino I do

And I'll consider that as I write the specs

@hyper-neutrino it seems only useful for if statements

okay ty

@lyxal i can't think of particularly reasonable uses for the other structures either

do you have something in mind; just ignore anything past the number of sensible branches and just merge the rest?

I just realised, probably <1b people have been rickrolled. What about people who watched it by choice?

so like (a|b|c) == (a|bc)

Also, does it count as rickrolling if you don't know what a rickroll is?

@hyper-neutrino ignoring extraneous branches is what I'm currently leaning towards

@hyper-neutrino that could be a possibility, but it leads to inconsistency with the multi branch if statement

@Ausername baby shark has 9B views and there aren't that many people on the planet (even counting all people ever alive at any point in the 5 years since then, i don't think we recycle people that quickly) so uh

clearly 1 view != 1 person

so even if every view is a rickroll

it wouldn't be 1B people

@lyxal fair enough

i mean i guess behavior around "invalid" syntax doesn't need to be particularly optimized since one could just like, not be bad

barring source layout or restricted source or whatever where one behavior or another isn't necessarily any more or less helpful as long as it doesn't error

@hyper-neutrino and besides, any useful functionality coming from optimising invalid syntax would have to be justifiable in regards to bytes saved/used

@hyper-neutrino Maybe I should start sharkrolling people instead

@Ausername no.

@lyxal true

postfix operators are modifiers right

and prefix are transformers?

Other way around

Postfix is transformer

Prefix is modifier

o

i assume we are keeping the latter for compatibility

will we be adding the former in the rewrite or will we add that as a feature for the future?

v2 will always have modifiers

v3 will have transformers

"Routine maintenance"? Come on...

This was not scheduled

And it's up again.

@hyper-neutrino we're implementing the v2 element list

poggers

meta.stackexchange.com/users/763699/user?tab=profile

Look at the communities list

yeah you can hide your communities from the list

O

@Ausername wait which sites

Why?

@hyper-neutrino MSE, about a minute ago

oh, okay

> This is a refactoring/rewrite, NOT a new language

BTW for v3 can we have Jyxal's behaviour of allowing multiple conditions in while loops?

@Ausername I'll consider putting that in 2.6

The way I'd implement it is if not pop(stack) break or something

whatchu mean by multiple conditions

At each |, popping the stack and if falsy, breaking

@Ausername I'll consider it all when I actually write the parser specs

huh.

sounds like it's just a subsititute for logical or?

Wdym?

it's just condition or condition or condition right

More condition and condition and condition.

so it's like having an infix and

sorta

so it kinda saves on stack manip

I guess it's cool

hey wait

where vybot

socket closed

and since for whatever reason my program doesn't crash when the socket closes

i can't just throw it into a bash while loop like i did for aplbot

welcome back vybot

@VyxalBot howdy!

i need to figure out a better idea eventually but for now

¯\_(ツ)_/¯

@VyxalBot hug

@VyxalBot lol u good

@lyxal I am doing you.

._.

!!/hug

ha, gottem

!!/hug

!!/hug

@hyper-neutrino ༼ つ ◕_◕ ༽つ

@lyxal ⊂((・▽・))⊃

me when i can't remember the command

L

impostor

@lyxal ⊂((・▽・))⊃

@lyxal ༼ つ ◕o◕ ༽つ

!!/repo list

@lyxal Vyxal | Vylight | Vyxal-Events | VyxalBotSE | Sandbox | vyxapedia

!!/issue VyxalBotSE Add sus command When you use `sus`, it returns the amogus unicode character feature request

Vyxal-Bot opened issue #25 in Vyxal/VyxalBotSE: Add sus command

wonderful

hyper-neutrino pushed a commit to VyxalBotSE/master: Implement github.com/Vyxal/VyxalBotSE/issues/25 - !!/sus

!!/pull

@hyper-neutrino pulling new changes; I will restart in a few seconds if any updates are available

!!/sus

@hyper-neutrino ඞ

!!/sus

@lyxal ඞ

:)

Issue #25 (Vyxal/VyxalBotSE) was closed by hyper-neutrino

chunkybanana opened issue #26 in Vyxal/VyxalBotSE: Close issue command

!!/hug

@Ausername ༼ つ ◕_◕ ༽つ

Issue #197 (Vyxal/Vyxal) was closed by chunkybanana

@lyxal nice. but... why is there a amogus Unicode char?

!!/sus

@Ausername ඞ

Why not

it's probably Javanese / Java

I was sleeping when it hit 1b views...

me too

CMQ: has anybody seen Rick astleys comment?

Yes

epic

Issue #160 (Vyxal/Vyxal) was edited by chunkybanana

Issue #177 (Vyxal/Vyxal) was closed by chunkybanana

Issue #160 (Vyxal/Vyxal) was edited by chunkybanana

Issue #160 (Vyxal/Vyxal) was edited by chunkybanana

chunkybanana opened issue #205 in Vyxal/Vyxal: Ǔ misvectorises

^ @lyxal Sorry

@lyxal why are you checking the string token 2 times? once elif head in "`»«" and then elif head == "`"

in lexer.py

line 156 and 184

Because backslashes escape backticks in only strings

uh?

oh okay understood

whats the difference between a tokentype.literal and a tokentype.string?

!!/run Ǔ

@PyGamer0 ⟨0⟩

Tams-Tams deleted branch VyxalBotSE/IWFSC-branch

^ forgot to do that

@math tt.literal is dead code

oh. why is it used, then?

eg. line 188

¹½⅓¼⅕⅙⅛²⅔⅖³¾⅗⅜⁴⅘⁵⅚⅝⁶⁷⅞⁸⁹⁰ⁿ <- @lyxal how about implementing this for Vyx

al

@math probably because I forgot to clean all instances of it up

@PyGamer0 because most of those aren't actually that useful for code golf

I'm probably not going to be able to help with the rewrite that much - not because I don't have the time, but because I'm pretty bad at understanding other people's code, and writing understandable code. But, I could make a start on debugging all the functions and creating new ones so we have those to come back to once the rewrite is done. After all, we're still going to be using the same functions for everything, the parser/compiler are just going to be completely rewritten.

^ Big chungus of text

@Ausername being able to understanding the code is something we're trying to make easy

That's why we have a style guide, comment structures, specs and planning documents

True

But then again, look at Jyxal's parser

I understand it because I made it

And I suppose it sorta makes sense

@Ausername why is Jyxal read only?

I archived it

yes why?

Gave up because I got to the stage where I had to implement a ton of builtins

and figure out how to vectorise generators

@Ausername no one ever truly knows how to do that

:p

@Ausername what's that?

@math a kind of in-joke now

Would you like an explanation of the joke?

Speaking of that, sorry

@lyxal yes

@math vyxal lore time!

alright

so

back in v0.x, if you had a python generator object, it was just converted to a list

e.g. itertools.power

because lists were implemented through a custom Stack class I had

ah

the major problem with that was that if you had a big generator, it'd take a long time to finish

so in v1, I made a Generator class that acts as a generic wrapper for all generators

And if you had an infinite generator

^ it wouldn't finish at all

@Ausername how can you do that in python?

def f():
    count = 0
    while 1:
        yield count
        count += 1

ah okay

of course

that's half the story

the second half is vectorisation

The approach for mapping/filtering over a generator is (correct me if I'm wrong on this) have a list of functions to apply to it when evaluated.

@Ausername no

just use map

ez

Oh

@lyxal anyhow

as you would know, python doesn't exactly vectorise things like addition/subtraction etc

like [1,2,3,4] + 4 doesn't return [5, 6, 7, 8]

^ that's what numpy is for, right?

it also doesn't vectorise my element functions

numpy is useless for the vectorisation rules vyxal requires

so instead, I just roll a custom vectorisation function

@lyxal element function what?

@math so you know how Vyxal is transpiled?

obviously I can't just transpile to straight python

so I need to define functions that implement the elements

(elements = built-ins)

you mean the list of functions in commands.py?

no

the functions in built-ins.py

wait a sec

github.com/Vyxal/Vyxal/blob/master/vyxal/builtins.py

i use numpy for the arrays so i dont have to handle Vectorisation

commands.py is a list of strings to transpile each instruction to

@PyGamer0 does numpy concatenate strings and numbers?

chances are it doesn't

oh that file

the helper thingy

I'll answer any questions on file structures/terminology once I'm done explaining the "vectorising generators" joke

Anyhow,

vectorisation mostly works fine on its own

as do generators

but the moment you mix the two together...

everything goes wrong

@lyxal Try it online!

because of?

github.com/Vyxal/Vyxal/releases/tag/v2.5.0

@lyxal wdym concatenate numbers

"abc" + 123

vyxal casts 123 to a string and joins

yay

Try it online!

Many commits have been dedicated to fixing vectorisation and generators

but none succeeded

correct

@lyxal Try it online!

we've even created a github issue tag for such purposes

@PyGamer0 well of course, but I can't exactly wrap in str when dealing with unknown stack objects

that's why I wrote a custom add function

and the vectorise function vectorises the add function

@lyxal cant you get the type

the thing is that it's not just addition that needs vectorisation

basically every function needs vectorisation

and numpy doesn't have functions with the same type overloads Vyxal requires

it ends up being easier to just have a vectorise function than to special case each comamnd

Jyxal's vectorisation has its own problems

Basically I need a way to declare scoped variables without hoisting

but that is hard

@Ausername did you archive it?

Ye

a while ago

alright, time to go write the parser specs

parser next?

correct

lexer first, then parser, then actual compiler

What's the difference between a lexer and a parser>

lexer = tokens

parser = grouped tokens

Oh

My parser was both

lol the current parser is also both

Vylight's isn't really a proper parser since it just needs to output a state

It has some parsing-related stuff, so you can tell when you're in a string, but by default numeric chars are number-coloured, function chars are function-coloured, modifier chars are modifier-coloured, etc.

Also, what happened to the static version?>

I just realized I've never used generators in Vyxal...

how does eg. Ḟ work?

wrong copy

one moment

!!/run ⁽+dk≈Ḟ,

[@lyxal: 58745697]
⟨0|1|1|2|3|5|8|13|21|34|55|89|144|233|377|610|987|1597|2584|4181|6765|10946|17711|28657|46368|75025|121393|196418|317811|514229|832040|1346269|2178309|3524578|5702887|9227465|14930352|24157817|39088169|63245986|102334155|165580141|267914296|433494437|701408733|1134903170|1836311903|2971215073|4807526976|7778742049|12586269025|20365011074|32951280099|53316291173|86267571272|139583862445|225851433717|365435296162|591286729879|956722026041|1548008755920|2504730781961|4052739537881|6557470319842|10610209857723|17167680177565|27777890035288|44945570212853|72723460248141|117669…

given a function and an initial vector, it repeatedly applies the function on a stack within the generator

roughly equivalent to:

def f():
    stack = initial
    for item in stack: yield item
    while True: # if you give it an extra number, it'll only generate a fixed number of items
        stack += function(stack)
        yield stack[-1]

Btw everything find-related is broken

probably

lyxal.pythonanywhere.com?flags=&code=%60hrkojo%60%5Co%E1%B8…

@lyxal why's d there?

to make the function have arity of 2

multiplying a lambda by a number changes it's arity

Oh yeah

that weird builtin

'night

Lyxal pushed a commit to Vyxal/fresh-beginnings: Create Parser.md

yes thank you vbyxak bot

Yay, parser

@user that's only the specs you sussy baka

!!/sus

@lyxal ඞ

^ is you

Bruh

!!/sus

@user ඞ

That's how sus works