Vyxal

emanresu A

00:08

Damnit

lyxal

just by a second too

emanresu A

Actually, does it?

Looking through everything using foldl, one moment

In theory, summing an empty list breaks.

user

It's not foldl's problem

If you want special behavior on empty lists, you should do it before calling foldl

0 is just an arbitrary default value

Steffan

0 is still the default

!!/prod

emanresu A

I know, I'm just saying that vy_sum should've been changed

> initial=None

Steffan

00:19

Look lower than inital=None

return working if working is not None else 0

emanresu A

Oh

Steffan

Although that is not the best approach

If we want initial to be None, that will still return 0

user

Yeah it should just be return working instead of returning 0 if it's None

Want me to change that?

Steffan

Yes, and set initial=0

user

What? No

Steffan

00:20

That was the original

user

That would break stuff

Steffan

Oh, nvm

I'm not thinking

Or wait

user

Of course not, you're Steffan :P

Steffan

vy_sum will need changed then too

can someone look at #973 before we do production?

user

Doing that rn

@Steffan Does it work on your computer? Too lazy to check myself

Steffan

00:22

~~too lazy to check myself~~

fine I'll try it

lyxal

works on my computer

user

Good enough for me

Steffan

there's a relelvant xkcd for that, but I'm too lazy to find it

Unrelated String

@user also i searched hoogle for Eq b => (a -> b) -> [a] -> [[a]] and didn't get any results in the base package; didn't actually know haskell had a groupBy :: (a -> a -> Bool) -> [a] -> [[a]] in Data.List :P

user

lol

Unrelated String

00:30

i guess what i implemented is equivalent to groupBy . (\f x y -> f x == f y)

this is your brain on golflangs

user

So do we close #975 and make a new PR that just does exponents?

emanresu A

CMQ: v2.11.3 or v2.12?

Steffan

That was crazy. I post a Vyxal answer, and literally instantly, and I'm not exaggerating, I see a +10 for that post

lmao

emanresu A

00:45

I'm divided between implementing tinylisp in Vyxal and implementing Vyxal in tinylisp

user

Do both!

Unrelated String

i think one of those would be considerably easier than the other

user

Yes, implementing Vyxal in tinylisp /s

Unrelated String

clearly

~~i should implement tinylisp in jelly~~

user

Too easy

Implement Jelly in tinylisp if you want a challenge

Unrelated String

00:48

oh yes

user

Bonus: Implement Jelly in tinylisp by implementing Python in tinylisp and running Jelly on top of that

emanresu A

Vyxal does in theory have tail-call optimization

Actually no it doesn't because stacks

user

CMP: github.com/Vyxal/Vyxal/issues/971#issuecomment-1126596413

@lyxal Thanks but don't approve so soon

The meat of it is still not implemented

lyxal

That's why I said part 1

user

Oh ok

I guess part 2 will be the first implementation, part 3 will be the realization that it fails horribly, part 4 will be a long struggle to get it right, part 5 will be a compromise that falls short of our goals, and part 6 will be some sketchy tweaking to get the tests to pass

Definitely not based on my own experiences

emanresu A

00:58

@VyxalBot T

user

T?

lyxal

@emanresuA all occurrences of one item in another, not what T does

user

CMP: Opinions on this? Would be an easy change to make

Vectorizing can easily be done with v anyway

a + a[0] doesn't seem to be as easy to do though

lyxal

problem is that I don't know how much backwards compatibility that'd break

user

Ah ok

emanresu A

01:07

@lyxal Oh

lyxal

@user it could be perfectly fine though

Unrelated String

@user and is probably just as if not more useful on general lists as on strings

emanresu A

Yeha

lyxal

@user so it's not a no, but a cautious yes

emanresu A

I've been wanting that for a while

Steffan

02:40

hey guess what guys, i found an ace. affects two elements

sympy

will post in security team later.

Unrelated String

nice

lyxal

another reason we need symvy lol

Steffan

this one has existed since at least v2.6

emanresu A

LMAO

Well, it's DLosc's fault /s

in The Nineteenth Byte, Nov 11, 2021 at 4:43, by DLosc

@lyxal Symbolic algebra /s

Steffan

0

A: Inject arbitrary code into a compiler (cops)

Vyxal, since v2.6 Why does Vyxal have so many ACEs? Commands with randomness aren’t allowed, and neither are EĖ†. Info is in the repo.

lyxal

02:49

Sometimes I really hate sympy for using eval

like real displeasure and annoyance at them

emanresu A

We could patch sympy

Steffan

...but it would take us a long time.

lyxal

there's been a few attempts at doing so, but they never got merged or progressed very far

Unrelated String

@lyxal wait does it actually

what the fuck

lyxal

yes

Unrelated String

02:51

wow

lyxal

github.com/sympy/sympy/issues/10805

github.com/sympy/sympy/pull/12524

emanresu A

github.com/sympy/sympy/issues/10805

ninja'd

Wow this is stupid

They're trying to build a safe subset of python

Why can't they just make a GODDAMN AST EVALUATOR

lwn.net/Articles/574215

lyxal

hey hey guys look at this

github.com/sympy/sympy/blob/…

that's the cause of all the headaches

that one simple function

expr = eval(
    code, global_dict, local_dict)  # take local objects in preference
return expr

and that's used by their expression parsing function

which is used by their sympify function

emanresu A

why

lyxal

Guess that means we really need to get all hands on deck to develop symvy

emanresu A

03:00

Honestly I have no idea how to implement my ideas for symvy

lyxal

Treat it as a pre-processor

Something that expands into a normal string you'd pass to Sympy

emanresu A

I meant this

lyxal

that's also what I meant

Steffan

04:28

One thing is, once we get symvy, then we have Vyxal 3, and we have to impelement Symvy in Scala...

Maybe we should write it in Scala from the start?

lyxal

but hopefully the scala symbolic algebra library doesn't do silly things like sympy does

maybe co-develop them so that they are developed in parallel

Alternatively, if scala links well with Java, we can just copilot the conversion of python to Java and use that

I say to Java because I know that copilot works well enough with python to java

But I don't know how well it does python to scala

So long as we use type annotations in symvy and not do any extreme dynamic type casting, we'll be fine for auto conversion imo

@user can you use Java libraries in Scala?

PyGamer0

@lyxal i think you can

Can I use Java libraries without limitation in scala or not? yes you can. Can anyone convert above code in scala? yes, I guess even you can with a little effort. See Java to Scala cheatsheet or A Scala Tutorial for Java Programmers. — senia Sep 24, 2013 at 8:32

lyxal

04:45

@emanresuA apart from the lists, I think I have an idea (or at least can formulate one once I have a few more examples of your idea in practice) of how to implement it

PyGamer0

@emanresuA i guess: fork ngn/k and make symvy :P

emanresu A

06:01

I have at least four tabs of this open

PyGamer0

07:35

@lyxal do you autogenerate the release notes?

PyGamer0

08:37

!!/info

Vyxal Bot

PyGamer0

!!/run 10 2 5ṅ

Vyxal Bot

@PyGamer0 0

PyGamer0

what is ṅ?

Vyxal Bot

09:30

[Discord Relay]
emanresu#3881: Well this is awkaward

lyxal

11:01

@emanresuA do you have any examples of how symvy lists work?

Because if so, I can get an understanding of them and start planning how to parse your idea

I'm thinking ANTLR at this stage

But depending on lists that might not be viable.

@PyGamer0 yes, github does it based on PR titles

@PyGamer0 insignificant, join on empty string or first positive integer where a function is true

PyGamer0

@lyxal oh

@lyxal and does sympy support lists?

lyxal

Maybe

I think it has array objects but not in algebraic expressions

PyGamer0

is that going to be a problem?

> not in algebraic expressions

emanresu A

@lyxal Pretty much like any other expression, certain operators can do stuff to them and they can't be compiled into sympy. Syntax is [1, 2, [3, 4, 5]]

lyxal

So what are they used for? Pre-processing?

How do they factor into the symbolic algebra?

emanresu A

11:33

They're more of an extension to the language to make it more powerful

They're not part of the symbolic algebra but they're nice to have and make turing-completeness a lot easier.

lyxal

But how do they get evaluated? And how do they make symvy TC?

PyGamer0

@lyxal presumably lists are evaluated like k..

lyxal

11:51

Aw dang I've just realised a major problem with any symvy initiative - Sympy won't return a string that is natively symvy

Meaning that any further Sympy work on that expression won't use symvy

What if symvy is a version 3 thing and make_expression just filters out quotation marks

That could be better maybe

PyGamer0

@lyxal oh.... yeah

lol

@emanresuA you should make your symvy idea into a separate language ..

lyxal

12:08

Actually, it shouldn't matter

No wait it does

Because either the output needs to be parsed back to symvy or symvy needs to be backwards compatible with Sympy

PyGamer0

just ditch sympy and make your own with mpmath /s

user

@lyxal on the jvm, yes, but not with js afaik :(

Btw I’m working on normal numbers in vyxals and we can extend it to symbolic algebra

lyxal

Third party library or custom?

user

You can use javascript libraries from scala js of course but idk if there’s stuff for that

@lyxal custom

lyxal

12:24

But for the algebra though

Seggan

14:22

@PyGamer0 theres a button

→ 70 messages moved to Trash

github will you please mark merged prs as "done" so i dont have to spam the done button in my notifs

lyxal

@Seggan how did you have the patience to select that many messages?

oop nvm

Seggan

i found that you can shift click messages after control clicking them

lyxal

ye

I just discovered that too lol

and to think I'm running for TNB RO lol /s

Seggan

argh cxx makes operator overloading so annoying

ima ditch operator overloading for my number type

lyxal

hey so when you make ANTLR grammers, what do you use/do?

I'm considering using ANTLR for symvy

Seggan

14:30

@lyxal ~~pay me 20 bucks and ill write it for you~~

what language is symvy gonna be in?

lyxal

python and maybe scala

Seggan

ok for python...

lyxal

I say maybe scala because version 3 go brrr

Seggan

lemme get my sources rq

ok so ull need to download the antlr jar

antlr.org/download.html

then to compile grammar files run java -jar antlr-whatever.jar -Dlanguage=Python3 -visitor -no-listener <grammar files>

itll generate a lexer, parser, and visitor in python

as for the grammar files themselves

lyxal

those are just EBNF right?

Seggan

14:35

whats EBNF

lyxal

Extended Bracker Nues (?) Form

it's a way of specifying language grammar

Seggan

idk

you can probably stick the lexing and parsing in one file

lexer rules start with an uppercase letter (i keep em all upper case) and parser rules with a lower case

you need to have a "root" parser rule that defines you file

usually file: program EOF;

basically tells antlr not to stop parsing midway through the file

the rest is relatively simple

antlr prefers matching as large a lexer token as possible

i.e.

TOKE
: ('A'-'Z')+
;
TOKEN
: 'A'
;

will always match TOKE

lyxal

so basically make sure definitions don't have much overlap?

Seggan

if 2 lexer rules match same length, the one defined first will win

@lyxal for lexers, yes

parser rules are context specific, as parsers should be

it wont match an if statement on a while

ok that didnt make much sense but im sure yk how lexers and parsers work

lyxal

mostly lol

Seggan

14:41

ok i gotta go have breakfast take a look at myxal's grammar for now

lyxal

well that's good because I gotta go sleep

Seggan

the lexer and parser are in 2 different files tho

lyxal

I'll look at it later today

o/

Steffan

16:26

@Seggan in python? we need it in scala, not python

@Seggan you can mark all as done at once

just select all and click done

oh ur talking about symvy

however i think we should use smth like antlr with scala too

hmm... github.com/sirthias/parboiled2

maybe I'll give that a go

Seggan

18:24

@Steffan why not antlr

they have js target too

user

18:48

@Steffan That is quite a bit more complicated than what we need and not too performant

As Seggan said, ANTLR should work. It's relatively simple, will basically be a single thing for all implementations of Vyxal (which is actually important now), and should work on all platforms

Steffan

19:12

antlr doesn't support scala

afaik

i guess we can just use java in scalla

user

19:42

Yeah, ANTLR supports Java and JS, so we can generate Java for the JVM version and generate JS for Scala.JS

emanresu A

20:06

@lyxal That is a bit of an issue... but I think it should be possible to parse sympy back to symvy

Vyxal Bot

Steffan153 created branch VyxalS/antlr

Steffan

21:56

i have no idea how we will make modifiers postfix with antlr

might have to do the element popping part from the interpreter

it doesn't support mutual left recursion

maybe seggan knkows

Seggan

just tried compiling my first myxal native program, g++ gives 1073 lines of errors smh

user

Oof

Seggan

22:20

@Steffan nope. idk. try SO or make em prefix

Steffan

what's SO?

5

oh stack overflow duh

yeah i googled and googled

ill just put that part in the interpreter or something, post-parsing

i tried chatting with copilot: gist.github.com/Steffan153/ca4d7f9242b02d12f45b9c848055c973

lyxal

23:40

Haha our repo has an ecin amount of stars

It's the funny xes number lol

Transcript for