Vyxal

A username

12:43 AM

What happened to static stuff?

hyper-neutrino

?

A username

In the new branch, all the static stuff is gone

hyper-neutrino

¯\_(ツ)_/¯

lyxal

1:25 AM

we'll put that in as we get more developed

Vyxal Bot

1:56 AM

hyper-neutrino pushed a commit to Vyxal/Fresh-Beginnings: add lexer spec

hyper-neutrino

@lyxal when you have the time, could you take a look at ^ and let me know what you think? I'm unsure if you like the idea of having few token types and having the parser determine what to do with structures later on, but I can extend the spec to make the tokenizer do extra work if you think it would make working with it easier

lyxal

@hyper-neutrino should there be tokens for LSQUARE, RSQUARE, LPAREN, RPAREN, LCURLY, RCURLY, SEMICOLON?

that could maybe make parsing somewhat easier

hyper-neutrino

I don't think those should be token types and the parser can just check if a token has type OTHER and value {

lyxal

okay then, sure

and how about function parameters

hyper-neutrino

I could make it do that, but from the way I usually do the lexer-parser part, it would probably be fine

lyxal

1:59 AM

should they have their own token?

hyper-neutrino

@lyxal oh like arities? or wdym

lyxal

@hyper-neutrino @name:1|...;

@name:varX:varY:varZ|...;

hyper-neutrino

oh

lyxal

@name:varX:4:varY|...;

that sort of thing

@name|code;
@name:number_of_arguments|code;
@name:variable|code;
@name:argument_list|code;
@name;

more generally ^

hyper-neutrino

wait, so what does @name:a:2:b|...; mean

lyxal

2:01 AM

pop into local variable a, pop 2 items into the function stack, pop into local variable b

hyper-neutrino

ah

so function parameters do not go into the stack itself

lyxal

variables don't

the numbers do

hyper-neutrino

or sorry i meant named variable params

what happens if we encounter non-alnums before the |?

for example:

@ab+cd+ef:gh-ij:5:kl×mn|...;

lyxal

ignore the sussy chars

hyper-neutrino

it would be easy to just allow those to be valid, since then function call and function reference can also just wait until they find a ;

lyxal

2:03 AM

that's what I meant

hyper-neutrino

okay

lyxal

strip the sussy chars

that's what happens now anyway

hyper-neutrino

oh get rid of them? okay

cool :+1:

and [A-Za-z_] is valid for variable names right?

lyxal

yes

hyper-neutrino

cool ty

lyxal

2:04 AM

30 secs ago, by lyxal

that's what happens now anyway

at least that's what happens with for loop vars

not so much with functions

but that's what i want

hyper-neutrino

mm okay

lyxal

Try it Online!

like that

hyper-neutrino

ah

right i forgot about the whole transpilation thing lol

lyxal

because that's a third, less explicit design goal of Vyxal: being good with restriced-source if need be

hyper-neutrino

lol

lyxal

2:06 AM

heck, for a good while, you could use any bracket to close any structure

Unrelated String

lmao

hyper-neutrino

lol

oh yeah what should that do

like encountering unopened close-statements

lyxal

!!/run 5 [`hey wait`|\this is sussy`)

Vyxal Bot

@lyxal hey wait

lyxal

@hyper-neutrino ignore them

hyper-neutrino

2:07 AM

olol

less error = better

good idea

!!/run 5 ]

!!/run ⟨1|2|3] 1+

@UnrelatedString Did you forget to put backticks around your code (`code`)? Remember to escape any backticks in your code (to type `hi`, enter `\`hi\``).

hyper-neutrino

2:07 AM

so | encountered where it wouldn't make sense should also just be ignored, I assume

Vyxal Bot

@lyxal ⟨2|3|4⟩

lyxal

@hyper-neutrino correct

Unrelated String

!!/run ]

hyper-neutrino

whee

Vyxal Bot

@UnrelatedString (output was empty)

lyxal

2:08 AM

yeah that seems broken currently

hyper-neutrino

thonk

lyxal

then again, the server is running 2.4.1

chances are I fixed a lot of this in 2.5.0/2.5.1/2.5.2/2.5.3

Unrelated String

what i'd say unmatched closing brackets should do might be to just auto-complete an opening bracket on the left end but that mgiht be a biiit not easy to implement parsing wise

hyper-neutrino

shouldn't be that bad although there are a lot of weird ambiguities we'd need to sort through

Aaron Miller

2:49 AM

What should something like λ[`false`|`true`; do?

lyxal

I would say "close the lambda and if-statement" but then one could make the argument to say "just let ; be a generic closer and have the ]}) close certain numbers of structures"

Vyxal Bot

3:46 AM

Lyxal pushed a commit to Vyxal/experimental: Yes

lyxal

3:57 AM

Fun fact: I'm currently writing a specification on structure types

hyper-neutrino

poggers

Vyxal Bot

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: Create Structures.md

lyxal

4:18 AM

Remind me to add a flag that makes lists truthy if any item is truth in v2.6.0

Vyxal Bot

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: Update Structures.md

hyper-neutrino

!!/issue Flag for alternate list truthiness behavior Flag proposal: lists are truthy if any item is truthy (as opposed to if they are non-empty) enhancement

Vyxal Bot

Vyxal-Bot opened issue #197 in Vyxal/Vyxal: Flag for alternate list truthiness behavior

lyxal

!!/run 10(n,)

Vyxal Bot

[@lyxal: 58707328]
STDERR:
[POST /execute] returned 500

lyxal

4:21 AM

oh

wow

oh

oh i see what's happening

hyper-neutrino

F

lyxal

the, uh, webhooks for auto code updating are still firing

and that's screwing up the flask app

hyper-neutrino

hmm

lyxal

because I had to rollback git

ez fix

just disable the hook for onw

!!/run 10(n,)

Vyxal Bot

[@lyxal: 58707349]
0
1
2
3
4
5
6
7
8
9

lyxal

4:26 AM

!!/run 6.9(n,)

Vyxal Bot

[@lyxal: 58707357]
0
1
2
3
4
5

lyxal

okay i see

good

Vyxal Bot

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: Update Structures.md

lyxal

!!/run 5 {:0=0=|n,1-}

Vyxal Bot

[@lyxal: 58707385]
0
0
0
0
0

lyxal

4:32 AM

hmmm

!!/run 1 [n,]

Vyxal Bot

@lyxal 0

lyxal

aha

interesting

hyper-neutrino

5:10 AM

@lyxal can you provide a complete list of situations where token grouping needs to happen? so far i have @...|, @...;, °...;, (...|, →A-Za-z_..., and ←A-Za-z_...

lyxal

that'll be what the structure specifications do

but i'm pretty sure that's all that needs name grouping

hyper-neutrino

ok ty

just need to know since the tokenizer needs to do that

should backticks within a name specification just be dropped out?

lyxal

depends

functions/for-loops? yes

variable names? well, the grouping stops there

hyper-neutrino

right

wait hold on

for loops don't need a variable name, do they

lyxal

correct

it's optional

hyper-neutrino

5:13 AM

er that's really annoying

A username

!!/run `lyxal`3(fKf∑

hyper-neutrino

because now not only does the lexer need to retain a state of what it's seen

Vyxal Bot

[@Ausername: 58707476]
llllllllllllllllllllylllllyllllllylllllllyllylllllyllyllllllyllylllllllyllyllllllllyllylllylllllyllylllyllllllyllylllylylllllyllylllylyllllllyllylllylylllllllyllylllylyllylllllyllylllylyllyllllllyllylllylyllylylllllyllylllylyllylyxlllllyllylllylyllylyxllllllyllylllylyllylyxlllllllyllylllylyllylyxllylllllyllylllylyllylyxllyllllllyllylllylyllylyxllylylllllyllylllylyllylyxllylyxlllllyllylllylyllylyxllylyxllllllyllylllylyllylyxllylyxlllllllyllylllylyllylyxllylyxllllllllyllylllylyllylyxllylyxlllylllllyllylllylyllylyxllylyxlllyllllllyllylllylyllylyxllylyxlllylylllllyllyllly…

hyper-neutrino

it also needs to scan all the way ahead and possibly understand the parse structure

lyxal

just concatenate everything in the variable name branch

hyper-neutrino

5:14 AM

wdym

lyxal

like (abc|...) will turn into (, a, b, c, |, ., . , ., ) right?

hyper-neutrino

okay so how do you handle (k|...)

lyxal

ooh

good point

hmm

k| would be it's own unit anyway

that would be as if you had no variable name

hyper-neutrino

oh

lyxal

so (, k|, ., ., ., )

hyper-neutrino

5:16 AM

so you can't do like (ak| 1 ←ak + ,) either

A username

Jyxal's lexer does that pretty much

lyxal

@hyper-neutrino (ak |...) maybe

hyper-neutrino

it might make more sense to just parse the code directly at this point

is the intended behavior of (ak|...) to be a variable-less for loop with k| as an element?

lyxal

simple solution: never assign | as a digraph

A username

wait what to do with V flag in loops?

hyper-neutrino

5:18 AM

that doesn't sound like that simple of a solution tbh but

like what is the intended behavior

A username

I guess just hope it doesn't happen

hyper-neutrino

the implementation details are somewhat less important than making sure whatever behavior is intended is seen

lyxal

between either treating it as k| or variable k, I prefer treating as variable k

hyper-neutrino

same. okay ty

but should k| be valid as a digraph in other places or should we just say that'll never happen?

and if so, what should the program like a b k| c d do

lyxal

@hyper-neutrino just say that will never happen

never use | as a digraph

hyper-neutrino

5:20 AM

okay

you also run into another problem

A username

The only place you need forward-references are (...| and @...|

hyper-neutrino

there is a codepage digraph that doesn't really assign a meaning it just grabs the character and gets it as a codepage index

wait actually that character isn't a valid variable name anyway so we can just not care

same with \ i suppose

lyxal

correct

hyper-neutrino

hm... this feels a bit peculiar but it'll probably work out

also does anyone even use named for loops or named functions

A username

no

lyxal

5:22 AM

not really

A username

I used a named function once

lyxal

@hyper-neutrino but it's a feature that makes this whole language unique

A username

true

hyper-neutrino

well ik i am just wondering

A username

Does Jelly have any sort of structures?

hyper-neutrino

5:24 AM

i am not in any way advocating removing them lol

@Ausername no

there are chain separators which cannot nest and postfix operators (which I believe lyxal was adding / has added)

jelly is strictly a golfing language and has no features similar to conventional practical languages

A username

Something like Jelly's ¡ would be nice for Vyxal

lyxal

that's just a for loop

A username

True

hyper-neutrino

not exactly

ntimes on a dyad will keep moving the previous iteration value to the other argument, essentially operating like a "fibonacci-like sequence"

1 ,3¡ 2 returns [[[1, 2], 1], [1, 2]]

lyxal

!!/run 1 1 5 (~+)W

hyper-neutrino

5:28 AM

the first iteration gives `[1, 2]

Vyxal Bot

@lyxal ⟨1|1|2|3|5|8|13⟩

hyper-neutrino

oh

well never mind then

Oh yes ~

what does ~ do

don't pop

5:29 AM

hm

!!/run 1 1 ~+W

@lyxal ⟨1|1|2⟩

!!/run 1 1+W

@lyxal ⟨2⟩

Don't pop for dyads, filter for monnads.

hyper-neutrino

5:30 AM

!!/run 1 1 5 (~λ2|+;) W

Vyxal Bot

@hyper-neutrino (output was empty)

A username

undefined for nilads and triads

lyxal

@hyper-neutrino v2.4.1 uses the old parsing scheme where it only took the next character

hyper-neutrino

did we decide what to do with λ(; and (λ), etc

A username

#1 is a cat lambda. #2 is a lambda in a loop

lyxal

5:37 AM

close every structure inside

A username

Character to change cop if you're interested

lyxal

5:49 AM

11 mins ago, by lyxal

close every structure inside

I'll finish the structure specs later tonight

Because I really should do some uni work now

!!/goodbye

Vyxal Bot

@lyxal o/

math

!!/ morning

Vyxal Bot

@math hello to you too!

math

oh nice didn't know that was a command

A username

!!/!!

!!/!!/flowey quote

!!/flowey quote

Vyxal Bot

5:59 AM

@Ausername Don't worry, my little monarch, my plan isn't regicide...this is SO much more interesting.

PyGamer0

!!/hellO

Vyxal Bot

@PyGamer0 hello to you too!

PyGamer0

!!/run h

Vyxal Bot

@PyGamer0 0

PyGamer0

!!/run 102910-29(10+[2|h](!)!)

Vyxal Bot

6:11 AM

@PyGamer0 58

PyGamer0

wow

A username

7:03 AM

VERY IMPORTANT: Random choice from an integer generates a random integer between 0 and said number, counting range-modification flags. Manual ranging costs bytes and doesn't finish within the lifetime of the universe for larger inputs..

math

7:29 AM

you could just scale the random choice, right?

Vyxal Bot

11:56 AM

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: Update Structures.md

lyxal

12:23 PM

!!/flowey quote

Vyxal Bot

@lyxal Hehehe... did you REALLY think you could defeat ME?

lyxal

dang

I really thought someone pinged me

but it was only vybot

Vyxal Bot

12:43 PM

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: I, Lyxal, have made a horrible joke that I'll probably end up regretting.

lyxal

!!/flowey quote

Vyxal Bot

@lyxal Hehehe... did you REALLY think you could defeat ME?

lyxal

there

it's balanced

!!/flowey quote

Vyxal Bot

@lyxal No...NO! This can't be happening! You...YOU! You IDIOT!

lyxal

what a troll

Vyxal Bot

12:51 PM

Lyxal pushed a commit to Vyxal/Fresh-Beginnings: Update Structures.md

lyxal

@hyper-neutrino I think the current version of Structures.md is mostly compelte

*complete

math

@lyxal could you help me a bit with gpt-3?

lyxal

what would you like help with?

@math ^

PyGamer0

1:16 PM

how useful are variables ?

i wanna know

(i need to see a cgcc answer)

!!/run 4|33$

Vyxal Bot

@PyGamer0 4

math

@lyxal i'm back so you use pipeline right?

lyxal

@math pipeline?

math

transformers

no?

lyxal

transformers?

math

1:19 PM

pytorch?

lyxal

for what?

oh

oh you mean for the AI stuff

math

gpt-3

yes

lyxal

they have an API

only OpenAI has gpt-3

they also have an online playground

where you stick a prompt in and it completes it

math

okay

lyxal

math

1:22 PM

wow

lyxal

math

nice, but unfortunately not true

lyxal

math

what's that?

lyxal

because you actually need an invite first

they need to accept your application

math

1:26 PM

oh okay

in that time I'll try out gpt-neo

lyxal

hence why I say it took a month for my access

also,

Once upon a time, there was a code golfer named Math. He was an expert at identifying the weak links in some of the most widely used algorithms. In one case, he was able to find a vulnerability in the encryption algorithm used to protect the Internet against rampant eavesdropping that resulted in the exposure of the private data of millions of people. When others tried to replicate his work, they found the same weakness but were not able to replicate his success. For a time, it seemed as though Math had found a “bug” in the algorithm.

Once upon a time, there was a code golfer named Math. is what I gave it

math

that's not me!

but well done, gpt-3

lyxal

!!/'night

Vyxal Bot

@lyxal o/

math

@lyxal wait

prompts.ai

is that really gpt-3?

oh nvm

lyxal

1:43 PM

It is, but it hooks up with your api key lol

It's basically just the OpenAI playground but with built-in dark mode

Anyhow

I'm off to sleep for real this time

!!/'night

Vyxal Bot

@lyxal o/

hyper-neutrino

2:20 PM

@lyxal how should mismatched structure closing tokens be handled when there is no structure to close?

For example, [@abc] should close the @... (with a ; presumably) but what about @abc]? should it close the @ and then drop the ] because it's unopened, or should it treat it as a regular non-identifier character and continue parsing the name?

user

I would suggest treating it as an implicit opening (@abc]foobar becomes [@abc]foobar)

Making it an error, making it part of the name, or skipping over it doesn't make sense

hyper-neutrino

so does (] complete to ([])? (the other way around would be confusing and hard to implement imo)

Razetime

like the keta challenge? i doubt it

user

@hyper-neutrino Hmm, that's an interesting one. It should complete to ([]) but what rules would the interpreter have to use for that?

hyper-neutrino

2:36 PM

wdym

user

Like, how would the interpreter know to use ([]) and not [(])?

Razetime

it has to close the brackets in a way that makes sense obviously

see question: keta bracket autocompletion

user

2:51 PM

lol it's by lyxal

13

Q: Keta Bracket Autocompletion

Special thanks to Bubbler and AdmBorkBork for supplying feedback about this challenge Keta (Keyboard tacit — clever naming, I know!) is a golfing language I’ve been working on recently. One of it’s features is it’s most probably unique bracket autocompletion system, which will hopefully allow f...

hyper-neutrino

3:25 PM

keta is the opposite of what I was proposing

user

3:56 PM

Oh

Aaron Miller

5:37 PM

Ok, who broke the site?

It's just spinning forever.

hyper-neutrino

!!/run kH

2021-07-25 17:45:51,057: Exception on /execute [POST]
Traceback (most recent call last):
  File "/usr/lib/python3.7/site-packages/flask/app.py", line 2446, in wsgi_app
    response = self.full_dispatch_request()
  File "/usr/lib/python3.7/site-packages/flask/app.py", line 1951, in full_dispatch_request
    rv = self.handle_user_exception(e)
  File "/home/Lyxal/.local/lib/python3.7/site-packages/flask_cors/extension.py", line 165, in wrapped_function
    return cors_after_request(app.make_response(f(*args, **kwargs)))

the bot is gone too lol

user

PythonAnywhere problems?

hyper-neutrino

!!/run kH

Vyxal Bot

[@hyper-neutrino: 58710599]
STDERR:
[POST /execute] returned 500

user

→ 2 messages moved to Trash

hyper-neutrino

8:16 PM

@lyxal would it be fine for function names / parameter names to auto-close / automatically move to the second part if they encounter invalid characters?

like make @ab×÷; be equivalent to @ab|×÷; and ab:c:23×÷; be equivalent to @ab:c:23|×÷;, etc

this is a reasonable way to make it work but I want to know what you think - i may just parse the code directly since it should theoretically work

it worked for proton (although debatably that can't be considered "working" in that case :P)

lyxal

11:04 PM

@hyper-neutrino ignore the ]

@hyper-neutrino hmmm, I don't know how I feel about that

My instinct is to say no because restricted source

As in, function names ignoring non alpha chars could be good for those kinds of challenges

hyper-neutrino

hm, fair enough

and I guess nobody would use those for golfing anyway

yeah I don't think I'll separate the tokenizer in that case

lyxal

!!/run kH

Vyxal Bot

@lyxal Hello, World!

user

11:58 PM

→ 4 messages moved to Trash

I think the welcome message should also include something like "type foobar to get a list of this bot's commands"

@hyper-neutrino ^

Transcript for