Great job, GitHub

I have a repository that GitHub claims contains SML files (it doesn't, afaik). When I click on the "Standard ML 0.4%" above, I get that first message because GitHub seems to have forgotten how to give itself queries

The url looks like https://github.com/search?q=repo%3A<reponame>++language%3AStandard+ML&type=code

Modifying the URL to use quotes around "Standard ML" makes that error message disappear

lmaooooooooooooooooooooooooooooooooooooooooo

Gah I forgot just how perpetually hot university is in the summer

I'm in a library right now and I feel like I might as well be outside

Gonna have to find a group of 5 to do an assignment with

5 people

Well, 4 non-lyxal people

riiiiiiiiiiiiiip

Well this is gonna be a big week

First I'm picking my classes for next year, then donating blood, then taking the SAT

With not one but two principal's- and superintendent's- advisory committee meetings

And that's in the first three days lol

nice

I've just started university and coordinating it with high school is pain

@emanresuA wait how u be in university and high school at same time?

it happens

i think it's some kind of arrangement between the school and the university for students who have run out of math classes and the like

Back in my day, there was this thing where 10th grade students could take a uni course along side all their usual high school work

No clue whether it could be used for credit or not because I didn't do it, but once a week some of the big brain kids in my grade would go to the uni and do a little learning

Subjects including some sort of maths, some sort of introduction to programming and some sort of literature study

Also probably history

Pretty much that yeah

I'm only doing one cours

Instead of uni, I decided to go down the rabbit hole of esolangs and now look where I am :p

Back then, I mean

I'm a full time uni student now lol

L

ratio + L + i have 13 more followers + grammarfail + F + scss + not even 100 stars + last commit 2 years ago + wth is that white mode

@Seggan I got the Gradle configure to succeed but now VSCode is screaming at me about missing script runtimes

seems to be a problem with the language server

looks like 2 versions of kotlin are being used for some reason

I think it's my language server breaking it

the errors look like Class 'kotlin.Unit' was compiled with an incompatible version of Kotlin. The binary version of its metadata is 1.8.0, expected version is 1.6.0.

I think this can be solved by updating the version of the server the extension uses

wish me luck! :p

I FIXED IT

:D

that almost never happens!

@Seggan ok, now there's just one error, for this bit:

tasks.generateGrammarSource {
    maxHeapSize = "128m"
    val path = File("$buildDir/generated-src/")
}

it says Unresolved reference: File

oh, I guess it changed to file (all lowercase) at some point

@RydwolfPrograms I'm working on a project in JavaScript and it's pretty nice to work with, ngl. Once you know where the janky bits are, you can mostly avoid getting tripped up by them. And the ability to just create an object and add properties as needed is really nice for rapid development.

If I had a magic wand, though, I would change how var, let, and const work (and turn on strict mode so you have to use one of them). Get rid of function scope and hoisting; all variables are block scoped starting at the point where they are declared. let should do what const currently does, except it should also prevent mutating the value. const should either do the same thing as (the new version of) let, or maybe do the same thing but only be allowed at top level.

And maybe function arguments could be immutable by default, and if you wanted them to be mutable you'd prefix var, like so: function popMultiple(var array, count)

@Ginger import java.io.File at the top

@Ginger yeah thatd be the most logical part

@Seggan fixed it by changing File to file

works fine

hrm

interesting

Well, my house is now in a Tom Scott video

uh, what

In the background

me when Tom Scott goes to Mars

@emanresuA This you? :P

lol no

Anyway, I assume it's actually this video, and that's cool (both the video and the fact that your house is in the background)!

@emanresuA wth that's epic

@DLosc Yeah, this is what I've been saying all along. In objective terms it's a really badly designed language, but when it comes to making things quickly, it's amazing

I'm aiming for a mix of the two with Cuprous; well designed and safe against weird pitfalls and runtime bugs, but also quick and simple to write

Ad hoc structs will be one of the features, for example

You can just make a struct and use it, and the compiler will do its best to make sure it's used sanely, and then later on you can add a more formal struct definition and stuff

maybe it's not a good idea to learn Kotlin by making a programming language in it...

eh, it's fine

On the contrary it's the perfect way IMO

Making a programming language requires learning a lot of different parts of a language

And it's really convenient to test

I have enough of Rabbit's syntax made for me to be able to start working on it... and I have absolutely no idea where to start

Lexer

I'm just kinda sitting here, paralyzed with existential dread

If in doubt, start tokenizing

@RydwolfPrograms already done *laughs in ANTLR*

If you already have the parser done and it's interpreted, then you're at the fun part

oh boy

You're doing a bytecode sorta thing right?

yes, rbt files get turned into brf (pronounced "boof") files, which are then interpreted by the interpreter

Will your bytecode be a register/stack/heap sorta architecture?

I'm not sure yet

That's what I would recommend

depends on what's easier for me to implement

Easiest would be no bytecode at all

why? it's not that much more of an complication

Yeah it is

For it to make any significant difference performance-wise, at least

hm fair enough

but I like the bytecode idea

@RydwolfPrograms how about I don't do bytecode now but I add it later?

That's what I'd recommend

Get it working, don't bikeshed/prematurely optimize with the bytecode

me when premature optimize

@RydwolfPrograms alright so: now what

Now you just...interpret it lol

Run through the AST and do what it says to do

You've done the hard part

I, uh, don't think it's actually that easy

like, how tf do I do typing? or GC?

Kotlin has GC right?

no clue, lemme check

When you write an interpreter in an interpreted language you get GC for free

noice

what about types?

What about them? You'll need to find a way to represent the data as data Kotlin can understand, I guess

hm

@emanresuA awesome

@Ginger youd have to rewrite a lot

@Ginger yep. jvm go brrr

@Ginger check the typing classes in rol

github.com/Seggan/Rol/blob/master/src/main/kotlin/io/github/…

@Ginger Most high-level languages have GCs. If you've written anything beyond Hello World in a language and memory management hasn't come up, it probably has a GC

But Cuprous will be a disproof of that :p

oh no

@user yep

Well, sort of. It will have refcounting (which some people count that as GC) and possibly a fallback GC for situations where borrowing/RC can't work

refcounting is gc

borrowing is not

what.

@mathcat can't see rn, what is it?

chess trainer ad

mmm

You have to witness it for yourself

It's a masterpiece

@Seggan for some reason gradle(?) is putting the generated ANTLR files in the base package instead of the antlr package

can you run tree?

part 2 is ... even better

.
├── build
│   ├── classes
│   │   ├── java
│   │   │   └── main
│   │   │       └── community
│   │   │           └── rto
│   │   │               └── ginger
│   │   │                   └── rabbit
│   │   │                       └── antlr
│   │   │                           ├── RabbitLexer$1.class
│   │   │                           ├── RabbitLexer.class
│   │   │                           ├── RabbitParser$Abstract_function_declarationContext.class
│   │   │                           ├── RabbitParser$Abstract_operator_declarationContext.class

@Seggan ^

> Aimchess - They put the "Based" in "Turn based strategy".

@Ginger i dont see that

Who is this Tom O'Regan and why has no one heard of him?

it putting all of them in community.rto.ginger.rabbit.antlr

@Seggan well of course now it starts working :/

if anything, try deleting the build folder first

@Seggan Yeah, technically, but I'm counting it separately since it's so much simpler

the GC ure talking about is tracing gc

just because its simpler doesnt mean it doesnt collect garbage runtimeingly

Yeah but it has none of the whole-program overhead of non-RC GC

And it doesn't allow for arbitrary data structures without fear of memory leaks, which I associate with GC

And if RC is GC then you could make a good case for borrowing

IMO it's a spectrum

Tracing GC is obviously GC, manual memory management obviously isn't

@RydwolfPrograms huh

it does

you still have to check refs thorought runtime

@RydwolfPrograms huh again

refcounting is the one i have a fear of memory leaks (circular refs)

@Seggan Yeah but only when that specific piece of data is cloned or dropped

@Seggan Exactly

RC doesn't free you from worrying about memleaks, while GC does

yeah i misread the sentence

(well, good GC, Node.js sometimes doesn't count :p)

@RydwolfPrograms if you dont have a borrowing system, you need to do this throught a programs lifetime. you cant exactly tell when a thing is gonna be dropped

ok, time to figure out how to structure the interpreter

@Seggan how does Rol do it?

I see things referencing "trees" and "nodes"

@Seggan Ah, well I'm thinking about reference counting the Rust way I guess

@Ginger thats the ast

Where the compiler can tell where specific instances of the RC pointer come into existence and are dropped

antlr produces parse trees, not asts

parse trees are horrible for doing stuff

so you use the RolVisitor class to turn it into ast?

yep

@RydwolfPrograms yeah in that case refcounting wouldnt exactly be truly gc, as you can just insert that into the compiled code

@Ginger RolVisitor also does some desugaring

then you have to resolve types with the ast

@Seggan how???

Certain recursive data structures would be optimized out without GC, like doubly- or circular- linked lists

hm, looks like the TypeResolver class is what does that

nope

its the TypeChecker

TypeResolver does dependency resolution

then why is it called TypeResolver

because it resolves types in dependencies/local code

uhhh

yeah i meant type checking, not type resolution in this

ohhhhhhhh

lmao my code looks... very similar to Rol's code

great minds think alike but fools rarely differ

@Ginger what part

yes

huh

@RydwolfPrograms Do note that if some object dies and it was the only referrer to a bunch of other objects, all those objects would be collected at once, possibly impacting your program's performance. That's fine for most things, but some applications need low latency and for that, a tracing GC that won't collect all the dead objects at once might be better

@Ginger You could also try using LLVM

hm

And use the Boehm GC, perhaps?

It'd be tougher to do, but a lot more fun (imo)

@user Oh that's an interesting thing to consider

Maybe I can have an exception where if there's a truly massive number of things referred to by a single object that gets dropped, it'd use GC

Or you could just have a stack of objects to delete

And if your deleting thingy takes up too much time, it puts all the stuff it needs to delete next on the stack and moves on to the rest of the program

Although one thing to consider is things that run on Drop

You might not want there to be a chance that the file handle or TCP connection you just dropped won't close until some time later

Although I guess typically you don't store enough open file handles or network connections in one object to cause noticeable lag when they're all dropped :p

How do runtimes with tracing GCs deal with that sort of thing?

I know Java no longer makes any guarantees that the finalize method will actually be called when an object is killed

Ew

What are you Ewing at?

The possibility of something being dropped and not taking care of what it needs to

That's usually put inside the close method in Java

When does that run?

There's an AutoCloseable interface, so you can do try (AutoCloseable whatever = ...) { ... } and at the end, whatever gets closed

So like Python's with?

Not really a fan of that personally

It'd be nice for things to clean up when they go out of scope, but most high-level languages don't seem to do that

@user what if you need it to stay alive out of scope?

@RydwolfPrograms yeah finalize is deprecated and used as a last resort

@Seggan That's a leak

The fact that you can close something and still pass it around is weird

no i mean say you return an open file

lol it's been so long since I've returned to CGCC

I never realised there was an on-topic TNB

@Seggan wdym

Oh, I see what you mean, I phrased that poorly

By "go out of scope", I mean like the end of their lifetime

@BetaDecay It's still more like "off-topic TNB" and "things totally unrelated to code golf" than "TNB" and "off-topic TNB" :p

So if you return an open file, that wouldn't trigger closing it, but if you do let _ = methodReturningOpenFile(), then you're dropping it, so it'd be closed

Oh and wb!

@RydwolfPrograms haha thanks

Is there a decent app for stackexchange chats yet

or just SE in general

No, progress has been going backwards. They deprecated the only existing app lol

lmao classic

> I know what you did (o_o)

Did VyxalBot tell you my secrets? >:(