The Nineteenth Byte

Seggan

15:00

@Bbrk24 Toyota sienna 2015

@RydwolfPrograms some do

mousetail

I don't actually write many libraries, mostly applications and they are normal GPL. I use LGPL for libraries

Rydwolf Programs

That's fine IMO ig. It's things like readline being GPL that makes me mad

Bbrk24

Swift uses a modified version of the Apache license iirc

I think it’s fine to crash if the username is EOF, right? Like presumably the user wanted to abort in that case

Rydwolf Programs

If Ginger has been killed and imitated by an AI, does that make them AGInger?

Ginger

15:17

ohhhhh baby I've got a race condition

:/

Rydwolf Programs

Exciting

Seggan

@Ginger rust go brr

Rydwolf Programs

FWIW I've had several race conditions in Rust

It's really easy to accidentally keep a mutex lock way longer than you thought you were/need to

And it's pretty annoying to figure out where that happened

Seggan

15:36

@RydwolfPrograms I think that’s a deadlock, no?

Rydwolf Programs

Oh right

Seggan

Tho I agree, I have to drop a lock manually in lightchat

I rather like how the memory system applies cleanly to resources

Rydwolf Programs

You're not making a chatbot library in Rust are you

Seggan

too bad this can’t be done for other langs

@RydwolfPrograms no

Rydwolf Programs

okay good

Seggan

15:37

lightchat, my chat client, is in rust

ive had to pretty much copy sechat into rust

Rydwolf Programs

You know I've already done that right

NPSP's in Rust

github.com/Radvylf/npsp2

While it's kind of cluttered and mixed with unrelated code, the chat handling is very robust and overengineered

Seggan

aaaaaaaaaaaa

But it’s integrated into npsp, not a standalone lib

Rydwolf Programs

Yeah, but you could probably have just copied and pasted a lot of it instead of having to reverse engineer sechat :p

Seggan

But then I would have to follow the mit license

while ginger game me the code with no conditions

Rydwolf Programs

What does that even mean

Just take the code and put it in your code lol

If the MIT license actually requires you to do anything, nobody actually does that thing

Seggan

15:41

And I have to copy your mit license into my projects too

Rydwolf Programs

?????

No you don't

That's just GPL

MIT is the second clearest way to communicate "do whatever you want with my code"

Seggan

Not as in I have to license my project as mit, just I have to display your license in my project too

Rydwolf Programs

I don't think that's the case, but to alleviate any concerns, I hereby give you permission to use my code without attribution or the inclusion of any stuff in your license file

Seggan

Too late :P

Rydwolf Programs

I've also been casually working on a true library

But it's not a high priority

mousetail

15:46

@Seggan Deadlock can be a kind of race condition

A race condition may cause a deadlock but not every deadlock is caused by a race condition

Bbrk24

16:04

You know what license we didn’t mention earlier

the WTFPL

> The license was confirmed as a GPL-compatible free software license by the Free Software Foundation, but its use is "not recommended".

2

hmm wait

is there an ANSI formatting code for strikethrough?

mousetail

I don't think so

Oh yes there is

code 9

But support is very limited

Bbrk24

I’m going to have to roll my own pseudo-NSAttributedString aren’t I

yeah it seems to only be (fully) supported on Apple platforms

I don’t know how NSAttributedString works underlyingly but storing a [Range<String.Index>: Set<Attribute>] seems like the most straightforward way to back it

actually I may need to implement a custom collection type because I’ll want the ranges in order, not ordered by their hashes

Ah no, swift-collections has an OrderedDictionary type, good

oh no that’s specified order not sorted order

mousetail

16:29

I'm enjoying hearing you suffer

Neil

17:01

@Bbrk24 what about the MPL?

Seggan

@Bbrk24 I thought you said you don’t have a mac

Bbrk24

@Neil I did mention that though

@Seggan Hence why I’m writing a TUI. Swift is supported for Linux, there’s just no graphics libraries because Swift on Linux is almost exclusively for server-side code

swift-collections, though maintained by Apple, is a cross-platform library

…if I want to support keyboard shortcuts, I’ll end up knee-deep in ioctls, won’t I

Seggan

@Bbrk24 heh, we’re all doing tuis today

@RydwolfPrograms on that note, mind quickly scanning my code and making sure I’m not blundering anywhere

github.com/Seaggn/lightchat

Bbrk24

I wrote a SIMD function to filter out nonfinite values the other day and even though it only ended up being like three lines of code I still feel good about having written it

Ginger

ugh

I'm going to have to rewrite a bunch of code because my jank has come back to bite me in the ass

Seggan

17:15

@Bbrk24 what’s smid

mousetail

SIMD = Single Instruction Multiple Data

It's a set of instruction on a CPU that let you do multiple operations for the price of one in some specific situations, which makes it fast

Bbrk24

single instruction, multiple data. By having large (256- or 512-bit) registers you can effectively do parallel processing on a single core

very common in graphics code, iinm

I’ve also seen a library that uses SIMD to validate UTF-8 in less than one instruction per byte

private static readonly Vector<long> _isFiniteMask = new(Enumerable.Repeat(0x7ff0_0000_0000_0000L, Vector<long>.Count).ToArray());

internal static Vector<double> MaskNonFinites(Vector<double> source)
{
  Vector<long> asIntegers = (Vector<long>)source;
  Vector<long> conds = Vector.Equals(_isFiniteMask, asIntegers & _isFiniteMask);
  return Vector.ConditionalSelect(conds, Vector<double>.Zero, source);
}

mousetail

17:31

@Seggan That repo is private FYI. If you shared it with Rydwolf and make it private on purpose ignore htis message

Ginger

no, he just spelled his name wrong

> Seaggn

4

mousetail

I'm seaggn right now

I feel seaggn could be the new poggn

bigyihsuan

17:51

sketched some example code out a bit: https://gist.github.com/bigyihsuan/8c8fd82cc0809e4e6a4fadb3d622404f

the `l :? T` is the "is `l` an instance of `T` or a child class of `T`" operator

user

It's going to be statically typed?

bigyihsuan

@user maybe yes maybe no, too early in planning for that

like

user

I personally would make it dynamically typed to make life easier

bigyihsuan

fields in class definitions need to know what type it is

user

And interpreted

If you make it statically typed and don't add generics, then you'll need to add casting and stuff

bigyihsuan

17:53

otherwise you have things like class Foo {bar,baz,quuz} and you don't really know what types that is

user

That's fine

This is a cursed esolang

bigyihsuan

mhm

designed to be terrible

user

(and languages like Python and Ruby don't force you to make the fields a certain type either even though they're supposedly "practical")

bigyihsuan

yeah...

i may go that route

but it'll be a headache when implementing the backend lol

(or make type annotations optional)

user

@bigyihsuan I feel like it'd be significantly easier to make an interpreter than to make a compiler

Type checking is quite hard

bigyihsuan

17:57

it was gonna be interpreted anyway because i dunno how to make a compiler haha

user

Yeah then dynamically typed will be much easier than statically typed

Well there won't be any "static" if you're interpreting anyway

(is it possible to have a statically typed, interpreted language?)

bigyihsuan

like i wanna use rust because i found a bunch of crates that really simplify making the frontend

user

Frontend as in GUI? I thought Rust didn't have too many crates for that

What crates did you find? I wanna start a Rust project of my own

Rydwolf Programs

I think they mean the interpreter frontend

user

Oh

bigyihsuan

17:59

yea

user

Like CLI?

bigyihsuan

lexing and parsing

user

Ohhh

bigyihsuan

(i also found a nice argument parser crate and i'm using that as well)

43

Q: Which languages are dynamically typed and compiled (and which are statically typed and interpreted)?

In my reading on dynamic and static typing, I keep coming up against the assumption that statically typed languages are compiled, while dynamically typed languages are interpreted. I know that in general this is true, but I'm interested in the exceptions. I'd really like someone to not only give...

user

Thanks

Seggan

18:18

@mousetail yeah I spelled it wrong

@user can confirm

@bigyihsuan kotlin + antlr

Ginger

rewrite finished

let's see if it works

narrator: it did not work

okay, it's drawing things now

progress!

Neil

18:33

@Bbrk24 hmm, well, I guess you did mention it, but nobody seemed to discuss it

Ginger

at least it's something

mousetail

Why is the name of PLDI server so long :angry:

Rydwolf Programs

Forgetting that you bought donuts last night and finding them in your backpack while getting ready for school is the best feeling ever

Ginger

^[[<35;23;30M^[[<35;23;31M^[[<35;23;32M^[[<35;24;32M^[[<35;24;33M^[[<35;25;33M^[[<35;26;34M^[[<35;26;35MTraceback (most recent call last):
  File "/home/ginger/.cache/pypoetry/virtualenvs/crosstalk-V9V4K-Ek-py3.11/lib/python3.11/site-packages/textual/message_pump.py", line 490, in _process_messages_loop
    await self._dispatch_message(message)
  File "/home/ginger/.cache/pypoetry/virtualenvs/crosstalk-V9V4K-Ek-py3.11/lib/python3.11/site-packages/textual/message_pump.py", line 544, in _dispatch_message

I've really gone and done it this time

Rydwolf Programs

Yikes

That doesn't look like a fun error

mousetail

18:45

Most of those lines seem inside the library

Ginger

I yielded a str instead of a Segment and all hell broke loose

good thing it was an easy fix

^[[<35;60;25M^[[<35;57;26M^[[<35;54;27M^[[<35;51;28M^[[<35;47;29M^[[<35;45;30M^[[<35;43;31M^[[<35;40;32MTraceback (most recent call last):
  File "/home/ginger/.cache/pypoetry/virtualenvs/crosstalk-V9V4K-Ek-py3.11/lib/python3.11/site-packages/textual/message_pump.py", line 490, in _process_messages_loop
    await self._dispatch_message(message)
  File "/home/ginger/.cache/pypoetry/virtualenvs/crosstalk-V9V4K-Ek-py3.11/lib/python3.11/site-packages/textual/message_pump.py", line 544, in _dispatch_message

yeah apparently if you do anything even slightly weird during rendering it shits its pants and spontaneously combusts

👍

mousetail

That's the disadvantage of dynamic typing

things can almost work for long enough that the source is hard to find

Rydwolf Programs

I feel like error handling is the least appreciated of the hard little things to get right, whether in language design (looking at you Rust), libraries (looking at myself), etc.

mousetail

For sure, I already found that for some reason when joining a room you don't have access to I don't get a 403 status code but just have to parse the text to find out

Ginger

uggggggh

my janky code isn't working

Bbrk24

19:09

@RydwolfPrograms I really like how Swift syntactically handles throw/try, but I wish it could be better-typed like in Java

Rather than a try block, you have a do block where the individual expressions that can throw are marked try

It’s also super easy to convert between a throwing function and one that returns a Result

(one way you have the get() method, and the other way you have init(catching:))

Ginger

AHA

I had two off by one errors :/

mousetail

Off by 2 error?

Ginger

no, two different off by one errors

Rydwolf Programs

An ooffff bbyy oonnee eerrrroorr

Seggan

@mousetail a deal breaking one for me

Ginger

19:25

oh boy I found another one

there are two sure things in software development: off by one errors

Ginger

19:44

progress!

Bbrk24

SAUCE

Ginger

S A U C E

Rydwolf Programs

𝒮 𝒜 𝒰 𝒞 ℰ

Ginger

20:04

𝕾𝕬𝖀𝕮𝕰

Ginger

20:14

well, after about half an hour of shoving + 1 into various parts of my code, I have ended up with this:

not bad, eh?

Seggan

21:00

I’d say so if you weren’t so far ahead of me

Ginger

~~now now, be a good sport~~

I did license sechat's code to you :p

caird coinheringaahin g

21:13

27

Q: We're regenerating all identicons

We've recently discovered a vulnerability in our identicon generation process. To remedy it, we have changed how we approach generating them and regenerated all identicons. We do not have any indication that any personally identifiable information (PII) was leaked as part of this vulnerability. S...

discussion profile-picture security gravatar identicon

Bbrk24

Depending on how work goes next week, I may end up making raw malloc and free calls from C#

Ginger

ಠ_ಠ

Bbrk24

C++ code may need to resize a buffer than a C# caller is giving it, but you can’t realloc a C# array

—-
there’s a chance I’m getting way ahead of myself, but if I wanted to use e.g. ctrl-I and ctrl-K to switch between chat rooms, how would I listen for that?

I know Ctrl-<letter> is possible to listen for because nano uses ^O for “save”

Ginger

21:32

I wouldn't know, Textual handles all that for me

Seggan

@Bbrk24 c#? I though u were using swift

Bbrk24

C# at work, Swift for personal stuff

Seggan

Ah

@Ginger yeah crossterm handles that for me too

Bbrk24

I have decided I’m doing all the rendering myself which means I will end up knee-deep in ioctls

Unless there’s a C framework I can use in which case I’ll gladly take it

but I haven’t looked for one

Seggan

21:50

Oooof

Would c++ do

Bbrk24

Maybe

Swift-C++ interop is far from complete but some things work

Like they only just added correct handling of virtual methods

Seggan

@Bbrk24 Have you tried (n)curses

Bbrk24

I’ll look into it

DLosc

@DarrenSmith One thing you might want to do is go to the access tab and add another user as a room owner, so they can moderate the room in your absence if necessary.

Seggan

@Bbrk24 another one: github.com/dankamongmen/notcurses

iirc ncurses is part of standard linux

Bbrk24

21:55

the annoying thing is going to be that anything that uses function pointers is going to be marginally compatible with swift closures

DLosc

@DarrenSmith I'm there with you. Part of the reason I've never quite learned popular languages like Jelly and Vyxal is there are just so many builtins.

Seggan

I wonder how well Guilia would perform

*Gulia

DLosc

@Seggan I've been hanging out in chat too much... I thought that was going to say dankamogusmemes

Neil

@cairdcoinheringaahing ugh, but thanks

Bbrk24

I’ll probably also have to install ncurses as a git submodule because of how SwiftPM works

But, one thing at a time. Let’s see if I can get login to work

DLosc

22:11

@DarrenSmith The Tarpit is a room specifically for esolangs and esolang development, although it's been frozen for a while due to inactivity.

lyxal

@RydwolfPrograms that's what I like about MIT. it's short, simple, lets people do whatever they want without having to spend time fussing over whether they're allowed to do so and still makes me no longer liable if my code fricks up their stuff :p

Besides, all the cool language designers use mit :p

Bbrk24

that’s a bold take considering Rust is Apache

lyxal

I mean golfing language designers

Charcoal, Jelly, 05ab1e, Vyxal, Nibbles, Cjam, SOGL and MATL

Basically a good portion of all of the biggest golfing languages ever

(languages like husk, stax and japt didn't have licences, so definitely none of this gpl stuff)

DLosc

22:59

@lyxal ~~You forgot Pip~~ :P

lyxal

@DLosc sheet I'm sorry

Ayo that makes you one of the cool kid golflang makers though :p

@DLosc that's what I like about MIT - it gives you the benefit of a licence and doesn't cause trouble for other people

Transcript for