The Nineteenth Byte

Aiden Chow

00:11

@lyxal very interesting news, what a coincidence that the name of the language is very similar to your username. off by only one letter, in fact.

emanresu A

Indeed, an extreme coincedence. How bizzare.

Yousername

00:33

Could I receive some more feedback on my sandbox question? codegolf.meta.stackexchange.com/a/25536

Rydwolf Programs

02:15

@lyxal Congrats!

Seggan

03:16

@lyxal hmm interesting

the name is very familiar but i cant pinpoint exactly where i heard it before

lyxal

06:05

LDQ: Say I have a list of items I want to fold by a function. And say that the function can actually be a list of functions. What should happen?

e.g. Should it cycle through the functions when folding?

Or should it consecutively map all functions and reduce by the last?

Adám

@lyxal That's what J does.

Unrelated String

i would honestly vote for cycling if this wasn't in a stack language yeah

but since it is i feel like the utility is reduced since you can just dump a list onto the stack for the main use case i can think of

i.e. you have a list of known length and have specific things to do with the elements

but true cycling could still be good for something like an alternating sum

lyxal

@UnrelatedString it would still be golfier though, right?

Unrelated String

wouldn't it be like (i can't find reduce lambda in the docs so i'm just going to pretend it's r) r+|-|*|/; versus ÷+-*/

lyxal

06:22

@UnrelatedString what if list length 5?

The lambda would cycle back to the +

Unrelated String

yeah as i mentioned it is still new functionality if you do need it to actually cycle

lyxal

@UnrelatedString Also it's not implemented at all yet :p

Unrelated String

:p

mousetail

08:07

LDQ When should closures compare equal?

user

The == operator should call both closures on every possible list of inputs and check if all of those results are true

Since that would either return false or never return, you can optimize that to just returning false

mousetail

Ah yes of course

user

So closures should never compare equal

Serious answer: try to make it so your type system doesn't allow comparing closures at all, because that doesn't make much sense

mousetail

Most languages seem to do it that way but you could just check if they are the same function and any bound variables are equal right?

user

You could check if they point to the exact same thing in memory ig

What is this for? What kind of language is it?

mousetail

08:13

Practlang

So in JS this is false:

let f = (a)=>()=>a;
f(2)==f(2)

But it seems like you could just cheeck if the bound variable a is equal

user

Makes sense that it would be false

When would you need to compare closures, though?

You could return an object that wraps around the closure, if it's something special

mousetail

A closure is already basically a object that wraps around a function

Unrelated String

08:36

you could i guess like
attach the ast to the closure value and substitute in the values of the closed-over variables
but that seems like overkill for something that would almost never come up

like seriously, make them impossible to compare is the only sane answer

mousetail

Comparing them is very useful though, even if it just checks identity. Eg. adding or removing a event listener

Comparing by AST seems problematic since you could make a minor change to one "copy" of a function and break a lot of logic that might depend on them being equal, or not equal

Unrelated String

maybe a less insane alternative would be, if a closure is represented as a function pointer and the closed-over values, just comparing all of those raw

especially since a use case like that would probably really not want it to be possible to have "the same" closure written two different places

mousetail

Yep that's my thoughts

I wonder why so few languages do it that way though, then I wouldn't need to useCallback every single function in react to make them compare equality properly

New Posts

10:25

0

Q: Should we update rules in faq about deleting answers?

In the faq it says that we need to delete published answers After some researches I found these two questions: Should I delete my sandbox answers after posting the challenge? We Have a Messy Sandbox where the community votes to delete the answers in sandbox after publishing: But in reality a...

discussion

New Posts

12:57

0

Q: Print all pandigital numbers

Given a base as input, output all pan-digital numbers. A number is pan-digital if it includes every digit in that base at least one. Every number is considered to contain infinite number of leading 0s. sequence rules apply. You may either: given a base and a index, print the n-th pan-digital numb...

code-golf number sequence

Rydwolf Programs

14:20

@mousetail Nah, removing an event listener by checking if the function's identical seems weird to me

Just either take or return an ID

Like setTimeout does, for example

mousetail

@RydwolfPrograms That requires much more administration

Rydwolf Programs

On the user's part or the library's?

mousetail

Users

Rydwolf Programs

I can't see how it'd be the former, and for the latter you literally just need a hashmap

mousetail

For the libraries point of view it's the same

Rydwolf Programs

14:21

@mousetail You'd already need to store the listener function somewhere

It requires exactly zero additional work

mousetail

It can be static though

Doesn't need to be a variable

Rydwolf Programs

Most langs allow immutable variables that you assign at runtime...

vals, lets without muts, consts, whatever your language calls them

mousetail

Only if you have some kind of object to store them in

A typical GUI application mostly has event listeners adding and removing other event listeners

There is no real way to manage this without global variables which is bad

Rydwolf Programs

Then make it so you specify an ID. That also means you don't need to store a bunch of lambdas in variables which is weird and ugly

mousetail

Those would be constants though

Rydwolf Programs

14:24

What would

mousetail

Not variables so no issue to store them globally since they can never change

Rydwolf Programs

Not what I mean

I'm talking about the code's readability not its best-practice-y-ness

mousetail

Nothing more readable than removeTimeout(updateScreen)

Ginger

^

mousetail

Literally nothing could possilbly be more readable

Rydwolf Programs

14:26

Yeah but what about const updateScreen = [big huge lambda]

That's very ugly

Ginger

what?

mousetail

I don't understand, big functions are a problem anywhere

You can define it as a normal function, it doesn't need to be a lambda

Rydwolf Programs

Only in languages where those are the same

mousetail

I don't know any languages that have global statics but no first class functions

(Or a way to define them yourself)

(Like a class with one method)

Rydwolf Programs

Rust differentiates between functions and lambdas (/closures)

Seggan

14:29

@mousetail just use hash(table|map)s for that

mousetail

There are traits that encompass both though

Seggan

@RydwolfPrograms i dont like that

mousetail

I imagine a event listener function would accept both

Rydwolf Programs

@Seggan Well it's for a very good reason

Closures capture their environment

Normal functions don't which is pog and good

mousetail

In fact, Bevy, a rust game engine, already does that and you can add and remove event listeners by identity

Seggan

14:30

@RydwolfPrograms maybe for a low level lang, sure

Rydwolf Programs

No, requiring you to deliberately make your function rely on external values is good

Pure functions are good

Seggan

but there shouldnt be a difference in a high level lang

@RydwolfPrograms yes

Rydwolf Programs

Yeah and this is a feature that makes pure functions much more the default

mousetail

Closures are basically requireed to make pure functions useful

Rydwolf Programs

@Seggan It has little to do with low vs. high level

Seggan

14:32

but i also dont want 2 lines of boilerplate to capture then like c++

2

Rydwolf Programs

Rust's closures are shorter than its normal functions :p

Also C++ sucks never insult Rust by bringing it up in the same conversation as it :p

Seggan

> C++ sucks

couldnt agree more

Rydwolf Programs

We should create a meta post to decide on a name for OTTNB given that it doesn't seem to be dying off

Permission to do that?

Seggan

still, i dont see the reason for distinguishing the types

Adám

@RydwolfPrograms What's wrong with OTTNB?

Seggan

14:35

most (function/closure) applications shouldnt care about purity

mousetail

^^

Rydwolf Programs

@Adám Idk, feels a little boring. I'd always imagined it having a name, and we've brought it up a few times.

Off-Topic TNB was kinda just a placeholder

Adám

TTB?

mousetail

@Seggan I can see the reason in languages where the exact memory something requires is important. But still you'd probably wrap it in some kind of wrapper like Box<dyn> so it doesn't matter normally

Rydwolf Programs

Thw Twentieth Byte is already the name of the mods' room

Adám

14:36

Oh.

Rydwolf Programs

And it's a little boring (sorry mods)

Seggan

@mousetail yes, thats why i was talking about low/high level langs

@RydwolfPrograms ~~well force them to rename it~~

Adám

I rather like OTTNB, as it can be (mis-)read in various fun ways, like Over The Top…

3

mousetail

In a low level lang you'd probably want the same available just not by default

Rydwolf Programs

One practical advantage of having a unique name is it's easier to move things to it

Remember the DBA incident? :p

Adám

14:39

How about Golf Overflow?

Rydwolf Programs

Well if we're proposing names maybe I should create the meta post :p

Adám

Yes, sure. Let's make a mini election process for the name.

mathcat

@RydwolfPrograms wasn't there a meta proposal to rename it?

Rydwolf Programs

There was one to "make it official" IIRC

mathcat

Although I might've confused it with LDW.

Rydwolf Programs

14:43

> Should we "officially recognize" it and add it to the room description for TNB?

Adám

I'd say so.

Rydwolf Programs

The answer to which was "yes", so we should probably add it to the room description

> General discussion for codegolf.stackexchange.com | Guidelines: cgcc-se.github.io/chatiquette | Off-Topic: chat.stackexchange.com/rooms/136286/off-topic-tnb

Looks good?

mousetail

Looks ok

mathcat

^

Adám

Maybe expand to "Use [link] for off-topic"

Seggan

14:45

^

Rydwolf Programs

How about "Go to [link] for off-topic discussion"?

"Use" feels vague

Adám

Sure, that works if it fits.

mousetail

Utilise

Rydwolf Programs

Actually "use ... for off-topic discussion" seems fine

"go to" feels like it'd ruin our code quality

2

mousetail

Maybe use "come from" like interecal?

Rydwolf Programs

14:47

room topic changed to The Nineteenth Byte: General discussion for codegolf.stackexchange.com | Guidelines: cgcc-se.github.io/chatiquette | Use chat.stackexchange.com/rooms/136286/off-topic-tnb for off-topic discussion (no tags)

3

So we can see how it looks

mousetail

Can you put line breaks in?

Rydwolf Programs

I don't think so

Adám

@RydwolfPrograms Looks fine, but technically not correct, as small off-topic forrays are ok.

mousetail

Barely any discussion here is on topic

mathcat

Update the chatiquette maybe?

Rydwolf Programs

14:50

Yeah, we should probably add it to the "be on-topic" section

I'm thinking while we're at it we should fork a very similar chatiquette for OTTNB

Adám

@mousetail This one is ;-)

Rydwolf Programs

Since it's now RO'd by the same people, it should probably have the same rules

chat.stackexchange.com/rooms/info/142818/…

You cannot include line breaks

mousetail

Sad

Ginger

@RydwolfPrograms FINALLY

New Posts

15:27

0

Q: What should we call Off-Topic TNB?

Since we've decided to officially recognize TNB's off-topic room Off-Topic TNB, it might be a good time to give it its own name. We'll do this the same way we decided on TNB's name; post one room name per answer. If you think the current name is the best option, feel free to create an answer with...

discussion chat

Adám

Q: What should we call Off-Topic TNB?

6

(↑ is because starring the NP announcement shows a non-descript url)

Rydwolf Programs

One problem is that TNB is already a pun on off-topic CGCC

Since the 19th hole is where you'd go after you're done golfing

So we should rename OTTNB to TNB and TNB to "discussion of the other 18 holes"

Seggan

17:12

> Rust has successfully embodied everything I want to avoid in a PL -- godawful syntax, bloated semantics, a fanatical community that makes poor rationalizations for it sucking as bad as it does, the ad hoc feel of the whole thing. If a language promising memory safety is infuriating (or, as they put it euphemistically, "has a steep learning curve") to the point that it makes you just want to do the memory management, yourself, you know it's a turd.

source

i disagree that it has a steep learning curve

ive only wrestled with the borrow checker once

syntax is quite nice too

but it could be a bit cleaner

Ginger

@Seggan that's impressively wrong, kudos to whoever wrote it q:

mousetail

The fan base is a bit overzealous yes, but that's true for basically any language

Seggan

id debate that point

rust fans are much more zealous about rust than any other lang ive seen (except lisps :P)

mousetail

Python fans can be pretty crazy too about denying the language's weak points

Seggan

tbh python doesnt have many weak points besides (imo) dynamic typing

Rydwolf Programs

17:21

That kind of points out Rust's surface level flaws while leaving out all the good stuff

Seggan

even significant whitespace is somewhat nice (id not add it in a new lang tho)

Rydwolf Programs

Rust does a lot of things very well

Its traits and structs and enums and stuff are a great alternative to traditional OOP, it keeps a lot of the cool stuff from JS (which makes it feel nice to use to me) while dropping the jank, its compiler is amazing at helping you figure out bugs, etc.

Seggan

yeah enums are real nice

though they could have been called differently

Rydwolf Programs

Rust's verbose syntax is the main real complaint in that paragraph, and that's a design choice

Not one I really like, but ¯\(o-o)/¯

Seggan

im planning to implement smth similar to rusts enums using union types in rol

Rydwolf Programs

17:25

Unions are just worse enums

Seggan

rol's enums will be your traditional java enums

@RydwolfPrograms but they are more versatile

Rydwolf Programs

How?

mousetail

In C you'd (almost) always combine your union with a enum, so it's basically the same thing with extra steps

Rydwolf Programs

I mean you can use unions to do stuff like turning a float into its binary representation, but that's cursed and unsafe

Seggan

@RydwolfPrograms can you say e.x. type ast = Expr | Int | String in rust enums

Rydwolf Programs

17:27

Yes

enum AST {
    Expr(Expr),
    Int(i64),
    String(str)
}

mousetail

Oh you mean typescript style unions not C style unions

Then yes I agree

Seggan

@RydwolfPrograms yes but you still have to match on it to extract the string

@mousetail yeah those

maybe i shouldve said sum types

Rydwolf Programs

@Seggan In what situation would you not need to do the same thing with your way?

How would you know it's a string and not an int unless you were already matching something else?

Seggan

as in i mean its still wrapped

you have to deconstruct it

i can just do node as String

sum types are also a good utility in general

Rydwolf Programs

Wouldn't that be a place to hide bugs?

Seggan

17:35

especially since rol doesnt have static overloading

@RydwolfPrograms not if im sure

by ur definition, casts are a place to hide bugs :P

Rydwolf Programs

@Seggan How are you sure?

You need some sort of enum thing at some point, why not just bundle the union with it Rust-style

@Seggan (and they often are :p)

Seggan

@RydwolfPrograms say if my function for some reason already knows the input node is a string?

@RydwolfPrograms i have java style enums

Rydwolf Programs

@Seggan Then why are you passing the sum type instead of just the string?

If anything it's good to discourage sum types in places where you know the type

Seggan

maybe because the superclass requires the sum type?

Rydwolf Programs

@Seggan Sure and I'm arguing you shouldn't :p

Seggan

17:41

@RydwolfPrograms agree

Rydwolf Programs

@Seggan Hm, that makes sense I guess. Still don't see how an if let or a match to get the string is any different from an as String in that situation

Seggan

shorter, a bit clearer, automagic failure if im dumb

Rydwolf Programs

And at least the if let/match makes the fact it can error explicit

Seggan

which would u rather see

var x = y as String or var x = if (y is String) ? y : throw TheCoderIsDumbError()?

Rydwolf Programs

Many Rust enums in libraries have .as_string()-like mathods btw

Seggan

17:53

not surprised

but wouldnt it be nice to have it as a language feature? ;)

Transcript for