Vyxal

lyxal

10:33

mfw updating mill breaks native because the corresponding helper library for native hasn't been published to maven yet

Bridget

[user] Lyxal

[user] I literally just reverted the mill update PR yesterday

lyxal

Well there's was a new one

0.12.1

Bridget

[user] The problem is spire being dumb

lyxal

You reverted the 0.12.0 one :p

Bridget

[user] It doesn't buuld for sn 0.5

lyxal

10:39

@Bridget no it's literally that mill is missing a helper library for compiling native

Bridget

[user] Yeah well don't let tge bot update mill for a bit

lyxal

repo1.maven.org/maven2/com/lihaoyi/…

Missing a 0.12.1 version

Tbh I'm thinking we drop native support

Bridget

[user] Yeah because mill is using sn 5 now

[user] 0.5 i mean

lyxal

What good has it ever done for us except make life painful

Bridget

[user] Hmm

[user] Does anyone use it?

[user] How much faster than jvm was it

lyxal

10:41

It restricts out regex, makes multi platform library support that much harder, doesn't even build on release anymore and people seem slow to publish libraries for updated versions

All so a very niche group of people can run vyxal 3 like somewhat faster

When everyone else uses the website because that's how most code golfers interact with it

Bridget

[user] Hmm

lyxal

It's the most tempremental build platform I've ever seen

And it's not pulling its weight

Bridget

[user] Okay i guess it does make our lufe a lot harder for little gain

[user] .aybe when it hits 1.0 we can come back to it

lyxal

Yeah

Early adoption isn't always the best thing

Bridget

[user] 👍

[user] Sometimes you gotta do late adoption

[user] Aka kidnapping

lyxal

10:44

It'll be so liberating to have regexes that aren't limited to some weird subset google made

The codebase can be terse again!

Bridget

[user] Lol

lyxal

You remember that hideous number regex (that granted we don't use any more because we dropped fastparse)? Well it was only hideous because native

Bridget

[user] Yeah i rememvrr *(unfortunately)

lyxal

Should probably remove fastparse from the dependencies lol

Bridget

[user] There's scala parser combinators in there for some reason?

lyxal

10:51

pfft

we haven't used that since the 3.0.0 pre releases lol

@Bridget 5 people at most

Bridget

[user] Ah ok

lyxal

compared to ~11 people using the jar

and who knows how many using the website

@Bridget ~~drop spire as well, roll our own complex number impl and use BigDecmial for arbitrary precision~~

Bridget

[user] Nooooo

lyxal

yesterday, by Ginger

it's just spending six seconds total adding and multiplying Spire's complex and real numbers

Bridget

[user] Man i set off to do thst once and had to givr up

lyxal

11:02

it is a little silly with perfomance

Bridget

@lyxal [user] Okau i believe we can figure this out

[user] Plus sieve

lyxal

dropping native also has the benefit of allowing symbolic algebra to be a feature

(separate thought to spire for a minute)

because I did some thinking, and it could work out if there was just jvm + js

use symja for jvm, and a js library for the js

wrapped in a shared class to abstract the different implementations

food for thought

and thought for food

Bridget

[noodle man jacob] Are you talking about removing support for it to run in native, so just have the web and JVM versions?

lyxal

yeah

don't have native as a release target anymore

Bridget

[noodle man jacob] Kai is written in Kotlin and I think they did the same, or at least partially dropped support, because JVM performance was way better

[noodle man jacob] Kap not Kai

lyxal

11:08

uiua brainrot

@Bridget it's not even performance

Bridget

[noodle man jacob] autocorrect moment

lyxal

it's because native is a right proper pain

27 mins ago, by lyxal

It restricts out regex, makes multi platform library support that much harder, doesn't even build on release anymore and people seem slow to publish libraries for updated versions

Bridget

[noodle man jacob] That too

[noodle man jacob] I saw yeah

[noodle man jacob] just saying that it's not a crazy decision, there is precedent

lyxal

the amount of times it's failed on macos, limited what we can do because three build reasons, and been annoying

not one of the things I like

Bridget

12:42

@lyxal [the cat] <a:yippee:1124707213815070782>

Bridget

19:37

[pacman] quick challenge idea:

print a non ascii character using only printable ascii. if your language uses a custom codepage, print a character not in the codepage

[pacman] is this trivial

[user] Good CMC

@Bridget [user] In many languages, yeah

[user] But good CMC

Bridget

20:37

@Bridget [RubenVerg] 257O?

[pacman] wdym 2570

[RubenVerg] no no the last one is a capital O

[RubenVerg] it's Vyxal 3 code

[pacman] we have 256 builtin

[pacman] then increment

[pacman] then O

[RubenVerg] Theseus doesn't suggest that

[pacman] theseus is wrong

[pacman] also uh

[pacman] 256 is not in the vyxal codepage

[pacman] so just use that

[RubenVerg] doesn't 0O work?

[RubenVerg] null is not in the codepage

[pacman] printable character

[pacman] ₇O

[pacman] this works

@Bridget [RubenVerg] unspecified

[pacman] i will amend that

Transcript for