Vyxal

user

00:03

I'd like to deny them

@lyxal You don't use groupImpl directly, you have to use group and that will call groupImpl (you can check out the number tests in InterpreterTests for an example of that)

Oh no, groupImpl is accessible in the entire package

It should go into an object

@Seggan (I did try to learn Racket once, and it seemed cool, but I just hated the IDE (Dr. Racket) and lost interest in the project itself (Wezl and I were going to make a fast Befunge (or a fungeoid, at least (is that how you spell it?))))

Seggan

clojure is real nice

user

It does seem nice, but I've heard the runtime errors are super weird because they're Java errors

And it has several different sorts of data type types to deal with the JVM

Seggan

@user you get used to them

@user i dont follow

you mean lists/vectors/seqs?

user

No like deftype, defrecord, and reify

idk maybe I just misunderstood their purpose when I learned about them

But you have like protocols and interfaces

And I just found it really confusing how to represent expressions for a language I was trying to parse

Seggan

@user ah

i never use em, too complicated, i agree

user

00:16

How do you define custom data types then?

Or is it okay to just use maps for everything?

Okay, reading the docs again, they're making more sense

Have you tried Typed Clojure?

Seggan

@user you dont :P

i use maps

@user no

but im interested

actually i dont think itll work for fig

i rely heavily on duck typing

user

But maybe for a different project?

Seggan

i dont use clojure outside of fig

user

Although adding types to a lisp would probably make it uglier than adding types to Python or JS (Python's type hints and TS don't look much different from Python without type hints and JS)

Seggan

true

user

00:20

I tried typed Racket once and it looked kinda ugly, not very lispy at all :(

I wonder how ugly typed Vyxal would be (I know it'd be a dumb thing to add to a stack-based lang, but just imagine)

Another digraph for type annotations, I guess

And the same design for classes that lyxal's come up with already

Darn it, I totally forgot to design the v3 interpreter with debugging in mind

lyxal

@user not very, I don't think

user

Yeah as long as the type system is super limited, it'd just be type hints and classes, just like Python

lyxal

and besides, there's some cases were static typing is impossible in stack languages

user

I think we may have to make a separate Debugger class with an entirely different implementation from Interpreter (for v3)

lyxal

so it'd have to just be hints

user

00:23

Yeah

Factor does have effects but not all functions can have a fixed arity or coarity

Would be interesting to see how cursed a typed stack-based lang would be

~~with gotos~~

lyxal

and then structures like the for-loop and while-loop wouldn't be able to tell you the type of the top of the stack after execution.

at least, not at compile time

user

Yeah, while loops would be nigh impossible to type

Seggan

@lyxal halting problem go brr

thats why the jvm requires typing annotations in the bytecode for loops

user

I kinda want to learn Guile Scheme actually

So I can use Guix

97.100.97.109

00:44

I wrote my first Vyxal answer (codegolf.stackexchange.com/a/257144/113573); feedback would be appreciated :)

lyxal

You might want to use vyxal.pythonanywhere.com as it runs the latest version

vyxapedia runs 2.4.1

which is from mid-2021

the latest version is 2.19.0 which was released 6 days ago

Try it Online!

other than that, I think it's a pretty good answer given Jelly is only 3 bytes shorter

97.100.97.109

I will do that, thanks :)

lyxal

also, 25 bytes if outputting a list of numbers/strings is acceptable

Try it Online!

and welcome to vyxal, we all hope you enjoy your stay!

user

01:17

Oh hey, it's the golfer formerly known as Adam

97.100.97.109

01:30

Yep, that's me, hey

On an unrelated note, I think there's an error in the documentation; the description for ℅ (Random Choice) in elements.md(github.com/Vyxal/Vyxal/blob/main/documents/knowledge/…) says that it outputs from 0 to n, but it appears it actually outputs from 1 to n (inclusive)...?

Vyxal Bot

Lyxal created branch Vyxal/fix-documentation-for-℅

Lyxal opened PR #1566 (Vyxal/Vyxal) (Vyxal:fix-documentation-for-℅ → Vyxal:main): Make `℅` state it's 1 to n, not 0 to n in the docs

emanresu A

should it be 0 to n?

Means you can do a 1 in n chance easily

Or generate a random inedx

lyxal

the idea between 1 to n is that it's cast to range first

but I'll open a pr for that too, and we can merge whichever behaviour we want and close the other

user

We can just use the flag for setting the start of a range, right?

Vyxal Bot

Lyxal created branch Vyxal/make-℅-from-0-to-n

user

01:41

Just to clarify, it's going to be 0 (inclusive) to n (exclusive), right?

Not inclusive on both ends?

lyxal

inclusive

that's how random.randint works

Vyxal Bot

Lyxal opened PR #1567 (Vyxal/Vyxal) (Vyxal:make-℅-from-0-to-n → Vyxal:main): Make ℅ go from 0 to n

lyxal

aight, #1566 is for keeping 1 to n behaviour, #1567 is for 0 to n behaviour

the endpoints of #1567 can be changed if needed

but this way, it's a matter of close one, merge the other

do not merge both #1566 and #1567

2

97.100.97.109

This is a silly question, but how do you take a list as input?

lyxal

01:59

Python style list

Vyxal syntax also works

Try it Online!

emanresu A

Also python tuple style

lyxal

@97.100.97.109 there's no such thing as a silly question here - we all have to start somewhere :p

97.100.97.109

Oh, I forgot the string quotes, whoops

Thanks

lyxal

@lyxal With the exception of the question "are cookies a good food item?". That's a silly question because cookies are obviously good food items

@97.100.97.109 happy to help :)

user

@lyxal No, no, it's a valid question, maybe they're referring to browser cookies. Not all of those are good for you

97.100.97.109

02:17

Also, the Vyxapedia tutorial says that the arity is optional and defaults to one (vyxapedia.hyper-neutrino.xyz/beginners#structures), but when I exclude the arity (vyxal.pythonanywhere.com/#WyIiLCIiLCLOu3wxOyIsIiIsIiJd) it tells me that "Arity must be parsable as an integer" (unless I'm doing something wrong)

user

Don't include the |

That's only necessary if you have an arity

vyxal.pythonanywhere.com/…

97.100.97.109

Ah, I see

emanresu A

03:00

And you can use T and d to change arity anyway

lyxal

11:56

Okay so a slight issue with the testing helpers - they don't compare floats very nicely

e.g.

ctx.push(VNum(5.1), VNum(-45.4))
        Interpreter.execute(AST.Command("+"))
        assertResult(VNum(-40.3))(ctx.pop())

Unrelated String

precision issue i assume?

lyxal

yeah

it's too precise

Unrelated String

could make some kind of float assertion that uses a tolerance, or could compute the result in scala for the assertion

the first is probably... way more generalizable

lyxal

well that's what I was going to suggest

vyxal 2 float equality already uses an episilon

so it'd just be porting that over

the main issue here is incorporating it into infrastructure

which is what my question was going to be

 Expected -2835860390359859/70368744177664, but got -22686883122878873/562949953421312 (ElementTests.scala:33)

That's the test error btw

the problem is that one is -40.30001 while the other is -40.30002

Unrelated String

you could hardcode the tests for vyxal equality, then actually run vyxal equality and assert that it gives the right result in all the other float tests :P

lyxal

12:09

@user can I run the test suite with mill?

./mill resolve jvm._ doesn't list any test commands

1

Q: Salacious Bacon Tripod

In a certain chatroom, we like making acronym jokes about the build tool we use called sbt. While it usually stands for "Scala Build Tool", we aim to come up with all sorts of meanings, such as: Stupid Brick Teeth Sussy Baka Training Shady Blue Tourists Seriously Big Toes Silly Bear Topics Salac...

lyxal

12:33

Ask for a tip about Vyxal. How is EOF detected? — lesobrod 2 mins ago

@lesobrod it's automatically detected

if you're asking about strings with newlines, just surround your string with " and include \n for newlines

if you're asking about how to tell when you're at the end of input, that isn't a thing, because input cycles back around to the first input

lyxal

13:46

I've tried wrapping the Floats in strings in the call to VNum to see if that makes things any better in the floating point test

because it looks like the inaccuracies are being introduced when the VNum is created

There should still be an epsilon comparison of floats though

But string wrapping should help maybe

Oh huh there's no VNum string overload

Coulda sworn there was

Well that's okay, it can be added in the conversions.

It'd be a simple call to Real

Monologging this now because bed is now

I'll see what y'all come up with in the morning

o/

user

16:13

@lyxal Real probably already has that in a method somewhere, you should be able to use that in VNum#equals

Okay looks like it doesn't lol

Maybe the method is in some typeclass but I can't find it

@lyxal Oh, I forgot to add a test module for the jvm

You can edit build.sc and copy the object test extends ... line into the jvm object

@lyxal lmao

(minor correction: It doesn't stand for Scala Build Tool anymore, it's just sbt now, not an acronym)

@lyxal jvm.test should work now

(in the branch we're were adding tests)

@lyxal There's VNum.from, I renamed it to apply just now

> y'all ever just see a meme and rinse orderly flower landings so hard

rofl

tmw you forget negative signs exist

user

16:43

@lyxal Done!

Not sure what epsilon to use though

Seggan

i read somewhere 1e-9 is standard or smth

user

Ah, thanks

Vyxal Bot

19:40

Aurathic forked Vyxal/Vyxal into Aurathic/Vyxal

user

New contributor?

97.100.97.109

21:02

It me

I'm just trying to clean up the documentation

But I see some inconsistencies... for one, it seems that kP and kQ don't work as advertised; it seems that kP includes spaces (in addition to some unprintable characters) and kQ doesn't output anything at all

It's also weirdly not in ASCII order

emanresu A

21:28

You're probably on 2.4.1

They work fine

lyxal

21:38

@user 1e-15 is what v2 uses

97.100.97.109

21:48

@emanresuA I'm using whatever version vyxal.pythonanywhere.com is at

user

@lyxal I'll change it to that later then

lyxal

!!/run kQ

Vyxal Bot

@lyxal ` !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_`abcdefghijklmnopqrstuvwxyz{|}~`

lyxal

The bot uses vyxal.pythonanywhere.com too

97.100.97.109

Hm, I have no idea what I did wrong

97.100.97.109

22:19

Okay, now it's working, I don't know what was wrong before

user

Gremlins. It's always the gremlins

emanresu A

You might've been using lyxal.pa before

97.100.97.109

22:43

Probably

Transcript for