The Nineteenth Byte

OldSandboxPosts

01:00

Sandbox posts last active a week ago: Erasure Poetry, Print every Unicode character (almost) code-golf unicode

mousetail

07:20

@lyxαl Could you please unstar all the message swith just one star?

Unrelated String

^

emanresu A

Done

mousetail

Thanks, Lyxal

emanresu A

It'd be nice if the starboard behaved like with certain discord bots - multiple stars are needed before it shows up

Unrelated String

ooh that would be very nice

in addition to cutting down on spam incidents like this it would also make it less committal to star something and see if anyone else wants to pile on

i guess that might also lower the threshold for what people feel is star-worthy though

mousetail

07:33

It could show up in the starboard for maximum 12 hours unless it has 2 or more stars

emanresu A

The star threshold has gone down a lot. Case in point, the bottom of the starboard is from a week ago

~~Back in my day~~ this time last year it was about two days

mousetail

I think it's better this way, messages get the attention they deserve

Unrelated String

up you mean then

^

emanresu A

Actually neither really

Just less messages

And definitely less interesting messages (cough cough several users whose names I won't mention)

Although maybe TST has pulled some of that as well

Unrelated String

@emanresuA also unless we're seeing different lists the bottom of the starboard is from more like three weeks ago (granted it's up specifically because it has 11 stars :P)

emanresu A

07:36

Smol screen

Unrelated String

:P

Simd

Is this sort of precise timing message new ? “ The community is reviewing whether to reopen this question as of 58 mins ago.”

lyxαl

@emanresuA yeah, tst has definitely taken up some of the starbait stuff I would have used to post here

Which is kinda a shame because I enjoyed this room being a bit more active

emanresu A

07:51

Agreed

lyxαl

This used to be a fun chat room. I think the whole cut down on noise thing was done a bit too seriously and drove away some of the culture that made this room somewhere I'd spend way too many hours in

17

You don't see nearly as many in jokes as there used to be

Which were what made TNB interesting and not just another main chatroom for a site

lyxαl

08:13

Like there used to be lore to this room, a story where anyone could join in and leave their mark. No one was deadly serious because that's the nature of the act of code golfing. If the site were about something professional, then a "professional" chat room may make sense, but code golf is about breaking the rules of conventional programming and pushing things to the extreme. TNB reflected that.

Now sure there were conversations that really probably should have been in their own room. But that's the thing - they retroactively can be, because new rooms can be made and messages can be moved

I miss 2021 TNB. I wish some of it could come back.

Unrelated String

yeah come to think of it it is weird that there's so much less tst activity than there used to be offtopic tnb activity

and i feel like tst actually was comparably active around when it launched

lyxαl

Because I think TST has turned into a general discussion room where being off topic in the 2021 TNB way would feel off topic there

Unrelated String

ooooh

yeah that's it

lyxαl

The silly room has become too sensible and now there's no ability to express silly anywhere

Which is a shame imo

Bubbler

saw TST and thought of T-Spin Triple

6

The Empty String Photographer

08:24

in Sandbox, Aug 4 at 5:38, by PetlinBOT

~ Nyot evewything is staw-wowthy...

lyxαl

But not nothing either :p

The Empty String Photographer

in Sandbox, Aug 2 at 9:56, by PetlinBOT

~ Stars breaks into your house get *smiles wemuvd walks away undew runs away p-p-p-peew-pwessuwe?!?! *breaks into your house

forest

09:17

@lyxαl Just make it silly again.

lyxαl

~~You know what? I will!~~

mousetail

Why make this room silly instead of TST?

lyxαl

Because it's what this room was built on. It's deeply enriched in transcript history

Silly is also a bit of the wrong word

mousetail

That doesn't seem like a good reason

lyxαl

And because this room has lost a lot of its charm and uniqueness. It used to be one of the freest rooms on the site, where you could always be more chill than other rooms

At least that's what I think, having been here when it was a lot different

forest

09:26

That's something that drew me to the room. I wasn't here in its heyday, but I enjoyed the tail end of the fun.

mousetail

Why can't another room replace that?

lyxαl

Because another room isn't the main room of the site.

And because no one really visits other rooms

mousetail

Maybe they don't visit because of the noise?

lyxαl

There's barely any noise there!

forest

It's not noise, it's not 4chan.[citation needed]

lyxαl

09:29

TNB could use a bit more conversation that's generally interesting but not strictly code golf

forest

Well you're the local mod, can't you make a decree?

lyxαl

TST should be the end destination for conversations, not the starting point. (most of the time)

@forest well you see I'm not

My site is PLDI

forest

Ah.

You're so active here I just assumed you were.

mousetail

I don't think there is a shortage of off topic conversations here

they still happen all the time

forest

Isn't this room smaller than it used to be too?

lyxαl

09:32

It is

mousetail

Yea but that's not due to overly harsh moderation

lyxαl

When the noise crackdown happened, everyone suddenly went all hush hush don't make any noise lest you disrupt the room

At least from my perspective

And that would have slowly driven away some people who would have made the room more active

forest

That explains it.

mousetail

Was this 2021?

lyxαl

10:26

Early-mid '21

mousetail

CMC Given an exponentially distributed random number as input, output a linearly distributed number between 0 and 1 based only on the input.

You can choose the parameters for the exponential distribution yourself

Ginger

11:36

@lyxαl yup

I joined right at the start of the crackdown and I felt pretty pushed to keep this room 100% serious all the time

hyper-neutrino

11:55

To be fair, there was a period of time where the room started falling to accomplish its primary purpose, and suppressing more fun conversations to make sure that people actually had a place for on topic discussions was a worthwhile tradeoff (IMO) at the time.

These days, it feels like there's simultaneously not much fun here while still somehow having to deal with noise all the time and on-topic discussions aren't the only thing that happen here even still.

I don't really have the energy to figure out how to approach this problem because all of my time is being consumed by a different community but I'd be open to rethinking the vision of this room, and like I mentioned, I think it's about time we switch from guidelines to formal rules to stop the problematic noise and maybe simultaneously bring back the community feel.

lyxαl

@hyper-neutrino I'm not saying it wasn't the right action to cut down the noise a bit. I just think it was taken a bit too far

mousetail

12:07

The noise nowadays is mostly all caused by a single person

lyxαl

@TheEmptyStringPhotographer problem is that's the kind of message that is cited as noise

hyper-neutrino

@lyxαl yeah, that's a fair assessment

it is a bit hard to cut down on noise without suppressing it a lot though (or I'm just bad at chat moderation which is probably true) but I do think the state of chat right now isn't the healthiest it could be (or has ever been, but at least it's not the worst)

mousetail

It's really hard to make objective rules about what is disruptive and what is just harmless sillyness. And if the rules are not objective people will be angry if they get consequences but others don't

lyxαl

@hyper-neutrino I don't think it needs to be suppressed to the extent where people change their behaviour completely to avoid even thinking of posting something. I think a culture of moving once there's a problem would work

UndoneStudios

@NewPosts I was trying to do this before I realized brute force was the only way I knew.

hyper-neutrino

12:16

@mousetail yeah that's true. to be clear, when I say strict rules I mean stuff like bots so there's a clear and objective line to be drawn

@lyxαl good point — didn't really think of that cuz unfortunately that's not really a thing I can do on discord where I have more experience :P but you're right; that'd definitely work better and would probably be the best approach

lyxαl

→ 13 messages moved to The Sand Trap

case in point about moving once there's a problem :p

missed one :p

UndoneStudios

@UndoneStudios if anyone knows a better way, I'd like to know

caird coinheringaahin g

13:05

@lyxαl Fwiw, this is why I opposed having an off-topic room for TNB. This room should be for people to enjoy, and is the most relaxed main room for the most "relaxed" main site. Unfortunately, it seemed like people had an attitude of "everything strictly on topic or everything is noise" back in 2022, and it just ended up hurting this room

lyxαl

I think it's good to have a default on-site place to move messages to, but I agree with you completely there

as I said:

4 hours ago, by lyxαl

TST should be the end destination for conversations, not the starting point. (most of the time)

Rydwolf Programs

13:17

@lyxαl To be honest I agree

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Pseudo-Quantum Bogosort Quantum Bogosort is as follows: Quantumly randomise the list, such that there is no way of knowing what order the list is in until it is observed. This will divide the universe into O(n!) universes; however, the division has no cost, as it happens constantly anyway. If t...

lyxαl

It feels like everyone has forgotten how to have a healthy balance of serious conversation and just natural day to day conversation

Like there used to be enough going on here that a TNB wikia could be made

Now it'd just be a single page

There's not much character in the room anymore because everyone has forgotten how to, in the words of caird, be relaxed in the relaxed room

I know I used to spend a lot of time thinking of goofy ahh messages and jokes to send to try and get a laugh out of y'all (expressed through stars)

Now I just feel like that isn't looked upon favourably at all

So I don't, and now I think it'd take me a while to get back into that flow of how it used to be

I think the unofficial code golf discord server is a good metric of how TNB could look like, allowing people to be less strict about off topicness

It doesn't have "noise" as such, but still manages to have serious conversations when needed and just whatever flavour suits otherwise

And even if there is noise here, that's why 3 extra ROs were added

Completely removing the fun defeats the whole point of the above :p

I think the solution is to stop using TST as a launching point for discussions

Or at the very least to start trying to bring discussions here

Instead of just losing them to a room that only a handful of people ever check

A room that also isn't that discoverable by new users either. TST is severely limited in the people a conversation can reach. It'd be better to have a chance for more people to engage and then move than to save like 10 seconds and not have many people see it at all

</rant>

noodle man

13:42

@lyxαl what is TST?

oh The Sand Trap

lyxαl

Yr

*ye

noodle man

i mean i firmly prefer Discord to SE chat

that's why i'm not very active here anymore

but i agree with what you say about it not being fun here anymore

lyxαl

And you only joined a year ago :p

TNB was way more wild in 2021

So if you think it's less fun after the time you've been here, imagine how much less fun it really is

noodle man

i joined TNB in like january i think

lyxαl

And it seems like it was even wilder in 2016-2018

noodle man

13:51

i think it mostly got worse in just the last two or three months

lyxαl

@noodleman your chat profile says you've been a chat member since July 2022

caird coinheringaahin g

@lyxαl TNB was lawless in 2015-2016 :P

noodle man

@lyxαl probably on SO

lyxαl

@cairdcoinheringaahing Aw man wish I was there to experience it :p

noodle man

i've been on SO since like 2020 but my first account got banned because they found out i was 12 lmao

lyxαl

13:53

Lol

caird coinheringaahin g

@lyxαl Have a read :P

48

Q: We need to talk about Chat

I would start this post with something along the lines of "without naming names", but I think it's time to pinpoint the issues we currently face in chat and take action. This is a follow-up to The state of chat. This post aims to identify specific problems and specific problematic users. Because...

lyxαl

I've read that before :p

I more mean wish I was there to participate :p

The transcripts make TNB look like the best chatroom ever

I mean, there's been some semblance of fun creeping back in lately, but it still feels like taboo

Nowhere near its full potential

noodle man

14:08

well i'm off to go make a better PFP for my account

New Posts

14:32

0

Q: Inverse trigonometric functions

There are 3 (commonly used) trigonometric functions sin cos and tan each of these functions has an inverse function You goal is to write 3 programs or functions (one for each inverse trigonometric function asin acos atan) that take a (real) number a input and return the result of the correspondin...

code-golf trigonometry

Rydwolf Programs

15:51

Okay std::io::Result can go die too

Why'd the Rust people have to ruin such a beautiful language with absolutely awful error handling norms

Ginger

17:28

@lyxαl ~~lyxal is a chaos demon confirmed~~

Rydwolf Programs

17:38

@Ginger Not me reading this as "cheese demon"

Ginger

that too

Seggan

18:25

@RydwolfPrograms how is it bad

Rydwolf Programs

It feels like you're completely blind

Seggan

I much enjoy it rather than exceptions

Bbrk24

Swift makes it very easy to convert between throw/catch and Result<T, Error>

Rydwolf Programs

The way Rust handles errors encourages bad documentation, makes it near impossible to handle different types of errors gracefully, and forces you to use gross dyn types if you want to be able to use ? conveniently

user

Me using anyhow::Error and unwrap() all over the place

Bbrk24

18:27

tbf, only difference in Swift between () -> Result<T, Error> and () throws -> T in the first place is calling convention

Seggan

@RydwolfPrograms I don't see the first point, second can be solved by impling From

user

@Seggan It gets really annoying if you have multiple kinds of errors that could be thrown (well, returned) from a function

Seggan

@RydwolfPrograms How do you propose to solve ur problems?

user

Monads are cool and all but it's really annoying to have to make a custom Error enum that wraps around io::Error and all the other kinds of errors your function can throw

Rydwolf Programs

@Seggan Go look up a Tokio or std::io function and tell me what different errors it can return

It's not only undocumented but impossible

user

18:29

@Seggan I think algebraic effects can help?

Rydwolf Programs

@Seggan I've mentioned my ideal way of doing it a few times

Seggan

@RydwolfPrograms io::Error?

@RydwolfPrograms if you did I forgot

Rydwolf Programs

@Seggan io::Error could be literally a million different things

Seggan

True

Bbrk24

Swift special-cases the handling of any Error for ABI reasons, which has its own set of pros and cons

Rydwolf Programs

18:30

Basically, there would be this special global enum of errors

user

Special global enum of errors sounds awful

Rydwolf Programs

Shut up and let me finish explaining this :p

user

:P

Rydwolf Programs

It would be namespaced, unlike a normal enum. So you could have Error::std::io::UnexpectedEof, for example

You'd register error types using some keyword

user

Oh, an open enum?

Rydwolf Programs

18:32

The compiler would then track every variant which a function can return

Bbrk24

Like TS union types :P

Rydwolf Programs

There would really be no global enum; ad hoc ones would be created

user

@Bbrk24 Yeah union types would be great, it's just that you kinda need subtyping then

Result<String, SpecificError | OtherSpecificError> would be nice

Rydwolf Programs

So you can match on the different error variants of the result a function returns

The Empty String Photographer

No activity at all => I browse questions => I come back = > Suddenly lots of activity

Rydwolf Programs

18:33

You'd have the normal ? operator, but also a way to specify a wrapper

mousetail

You'd essentially get a Java style declare all possible exceptions system

Bbrk24

So you know how normally impl T and dyn T aren't interchangeable? Swift special-cases the ~~trait~~ protocol Error so that it can do that

user

^^

I unironically like Java's throws

Rydwolf Programs

@mousetail Ew nasty

No, this would be done by the compiler, not written out explicitly

mousetail

It's what you are describing

Bbrk24

18:34

Java's throws would be better if you could say "throws like <other function>"

user

Like Swift?

That would be cool

mousetail

So your function signature can change without you noticing?

That sounds like bad design

Bbrk24

Swift error handling is weird

Seggan

@Bbrk24 what's the like

also should we move to TGC?

Bbrk24

There are some really clever things about it and some really annoying things about it

Rydwolf Programs

18:34

@mousetail Technically yes

user

@mousetail This is how type inference works too so /shrug

mousetail

Rust doesn't allow type interference on function signatures

Bbrk24

@Seggan [prepares textwall about swift error handling]

Seggan

@TheEmptyStringPhotographer programming language debates are always full of activity :P

user

@Bbrk24 Dew it (seriously) :P

The Empty String Photographer

18:36

Secretly modifies rust compiler No more errors, as you requested.

mousetail

Except for lifetimes but even that is derivable from just the declaration

user

Make a question on PLDI and self-answer it

Seggan

@TheEmptyStringPhotographer that's an even worse scenario

user

@mousetail Langs like Haskell do let you do that, though, and it makes refactoring easier

If one function changes, you don't have to go around changing every other function's signature

mousetail

You'd need to change their content to catch the extra error though

Rydwolf Programs

18:37

Good!

Currently, it can still change, it's just an additional enum variant you never check

mousetail

No not good, a breaking change should require a signature change

Rydwolf Programs

That's already not the case

This just moves in that direction

user

The signature would still change, just not explicitly

Rydwolf Programs

Everyone just does dyn Error or io::Error

Aiden Chow

@mousetail completely unrelated to current discussion but i remember ur bday was around this time of year... idk where i got that impression from lol

user

18:38

The autogenerated docs would contain the signature

mousetail

@AidenChow Yea it was friday

Aiden Chow

@mousetail oh shit actually, happy belated bday

Rydwolf Programs

If people actually used the most specific error available and documented them properly instead of making giant fucking nasty enums of errors where variants are shared between functions, this wouldn't be necessary

But Tokio and even Rust's own stdlib shows people can't be trusted to do that

user

@mousetail Happy birthday! (not saying belated birthday because I'm actually nearly a year early, not a week late)

Seggan

~~lets just all use algebraic effects and be done with it~~ /srs I don't think algebraic effects are a completely good error model

Aiden Chow

18:39

oh i remember where i remember it from now, it was ur radiation hardening challenge u put on the leaderboard one day it was ur bday lmfao

Rydwolf Programs

Altho one thing I didn't consider when coming up with this was handling different OSes

A big part of why io::Error is like it is is handling different OSes

mousetail

I think the main issue with Java error handling is not that you need to list them manually but more so that the errors are all not really specific enough

Rydwolf Programs

@RydwolfPrograms And IMO the way to fix that is creating a hard boundary between OS-specific code, and code that interfaces with an abstracted virtual OS

mousetail

WASI-only language?

Rydwolf Programs

what?

mousetail

18:43

An abstract OS independent IO layer

Rydwolf Programs

idk

user

@RydwolfPrograms Can't you have an enum OSSpecificError in that case?

mousetail

You probably want a high level error enum with a property or field that allows you to get the underlying OS specific error if you really need it

Rydwolf Programs

Noooooooo

The Empty String Photographer

@RydwolfPrograms What Happeneddddddddd?

Rydwolf Programs

18:45

IMO there should be two sets of functions: one for working at an explicitly OS-specific layer, and one that's abstracted over any OS

Trying to do both at once is like trying to bake your dessert into your entree

user

enum OSSpecificErro {Windows(WindowsError), MacOS(MacOSError), ...}

Seggan

what about SeggOSError?

user

Ah yeah, adding OSes would be problamtic

mousetail

Probably just need WindowsError and PostxError

Seggan

what if SeggOS is not POSIX-compliant?

mousetail

18:49

That's very rare, most other OSs, while not 100% complient, at least use the same error codes

Even the first few basic error codes on windows are the same as on POSIX

Bbrk24

Swift has a protocol in the standard library:

```swift
public protocol Error: Sendable {}
```

However, it's full of compiler magic that can't be inferred from that declaration alone.

Swift has protocol existential types spelled `any P`. For historical reasons, in Swift 5, some existentials can just be spelled `P` instead of `any P`. Most of these are laid out in this five-word box:

```
+--------+--------+--------+--------+--------+
| value or pointer | vtable | wtable |
+--------+--------+--------+--------+--------+

If typed throws is added in the future, all existing stdlib functions that throw must be rewritten as throws (any Error), due to ABI stability.

Swift also has the keyword `rethrows`, which is pretty cool. A function marked as `rethrows` must meet the following requirements:

- It has at least one function-typed argument with `throws` in its type signature
- It will throw an error if and only if it calls the function-typed argument and that call throws. In 99% of all cases, it simply propagates the error unchanged, but it's allowed to catch that and throw a different error.

Some obvious candidates for rethrows are all the functional methods on Array, like map, filter, etc

Aiden Chow

19:27

@Bubbler ayo fellow tetris player, nice

DLosc

19:51

@lyxαl ~~Memes~~ TNB used to be funny

lyxαl

23:37

Thanks word

very cool

bro think it's code golfing

and again, trying to use who knows what golflang

ïĥł is the letter sequence

Transcript for