The Nineteenth Byte

The_AH

13:00

Yes...

guest4308

wait, was just chrome://colors ever a thing that worked on its own on your version?

I think you were using chrome://color/red and such, right?

The_AH

It was chrome://red.

guest4308

ah

The_AH

And some colors were nested inside chrome://colors/hidden.

It did work on it's own on my version.

guest4308

idk where I got the singular from then

does just chrome://red work?

The_AH

13:02

No.

guest4308

then sounds like that flag prob enables colors for some reason

wonder why?

and what else it does

The_AH

That flag isn't related to certificate authorities then.

I am not sure why it does that and what else it does.

Wait try opening a Chrome session with that command line flag added.

@Ginger This is crazy.

Ginger

what do you want

The_AH

So this --ext-ca-run flag enables colors.

Try it on your Chromebook.

Ginger

actually no don't answer: you had no reason to ping me like that

The_AH

13:06

Well sorry.

Ginger

I don't want apologies, I want changes in behavior

Simd

@UndoneStudios no. Are you thinking of writing an answer?

Hi @mousetail

Ginger

oh, that reminds me: @guest4308 welcome to TNB!

guest4308

thanks

@The_AH no change, but going to chrome://version shows that the flag didn't actually get added

mousetail

@Simd Can we assume 2^-8*n error rate for miller-robin primality test?

So we need only 4 iterations to get to the required 2^30 bound

Neil

13:15

@lyxal actually chrome: was a protocol in Firefox's predecessor, Netscape 6, back in 2000.

it means something completely different in Firefox though

Ginger

*cough*

Neil

it used to be the protocol used by extension pages in Firefox 52 and earlier

UndoneStudios

13:31

@Simd was

Simd

@mousetail isn't it 4^-k ?

For k rounds

UndoneStudios

my factorizing algorithm was coughy

Simd

1

A: Miller Rabin primality accuracy

A more rigorous statement is as follows. Suppose we want to consider whether $n$ is prime. Let $W_n(b)$ be the statement that $1 \leq b < n$, $b^{n-1} \not \equiv 1 \bmod n$, or there is some $i$ with $2^i \mid (n-1)$ and $1 < \gcd(b^{(n-1)/2^i} - 1, n) < n.$ If $W_n(b)$, then we say that $...

@UndoneStudios what does that mean??

UndoneStudios

on the surface, it looks like it should work

it really broke down on running

mousetail

@Simd It's -4^-k in the worst case, -8 in the average case

UndoneStudios

13:36

@Simd probably not the right descriptor. I'd like to amend that to deceptive

Simd

@mousetail ah....

Can you give a reference for that?

The_AH

@guest4308 Hmm...

Simd

@UndoneStudios :)

The_AH

Not sure how to continue with this.

mousetail

I read it on wikipedia en.wikipedia.org/wiki/Miller%E2%80%93Rabin_primality_test

The_AH

13:39

Wait if you add invalid flags it doesn't show them.

mousetail

Wikipedia cites this paper: math.dartmouth.edu/~carlp/PDF/paper88.pdf

UndoneStudios

wait i fixed it

I was adding to the factorization list the factorization of n%i instead of n//i

The_AH

That means in your system it's an invalid flag.

This could be OS-specific then.

Simd

@mousetail I think I have to say no. Sorry

The_AH

Just for a screenshot...

mousetail

13:40

Ok, understandable

Simd

I was wondering about a challenge for factoring 64 bit numbers

The_AH

Simd

Does this interest/ scare anyone?

UndoneStudios

Yes

Simd

@mousetail what language are you thinking of doing it in?

I am always hoping for Julia

mousetail

13:44

Rust, with fixed with integers

I hope they'll be faster than vector based integers everyone else is using

UndoneStudios

I wonder if there exists a number such that doing a recursive factorization would lead to a RecursionError in Python

mousetail

but probably will be much slower since I'm implementing my own primality check etc

@UndoneStudios For sure yes

UndoneStudios

@mousetail I mean what's the value

should be rather big

mousetail

1000! will for sure

UndoneStudios

of course it depends on the recursion limit and the number of prime factors

@mousetail by factorization i meant prime factorization

mousetail

13:46

It should still work

You will have at least 1000 prime factors when multiplying 1000 numbers

UndoneStudios

What's the smallest such number?

mousetail

Though some will appear multiple times

2^1000 should be the smallest one

Assuming 1000 recursion limit

UndoneStudios

Yeah that's correct

Assuming defaults

What about unique prime factors?

mousetail

Well then it would just be the product of the first 1000 primes

UndoneStudios

I figured as much

39180681289998166862525016396412762092247944781178736447765210657937332632759741323703903297578964643192459118230872038298379059450430533456584565981705298046725462784808035878284021916522787269918942967514393443344486820134237032455322266270384977697979829784314376617359793750274879038677807936189811099500772814212067673173321367078402023271835800079808990130406599499085971986269339629610636948161162415782251820

Factorization of this number may result in a RecursionError

Simd

13:56

@mousetail that sounds very cool

@mousetail it would be interesting to profile the C++ to see if it is spending most of the time in the primality test

mousetail

C uses GMP which is very highly optimized, probably way better than what I can achieve

Simd

@mousetail I don't think C is any faster than rust

mousetail

No, but GMP is faster than what I can write

Simd

At least it's possible to make rust as fast as C I think

@mousetail don't doubt yourself!

Fortran used to be better for high speed code

I don't know if that is still true

mousetail

For sure a highly optimizes rust code can beat C but I don't have the patience to micro optimize

Simd

14:03

I am no expert but it can be hard to optimize at a low level unless you really understand what GCC or clang is going to do with it. They can be quite different from each other

This is particularly true for simd code

mousetail

For sure, it takes a lot of trial and error

GMP has been trying different methods for decades

Simd

@mousetail there should be community answers where everyone is free to improve the code

mousetail

People do sometimes comment suggestions for how to improve things

but people don't generally try to run the code

Simd

I like it when that happens

mousetail

GMP gets like actual PHD students writing their dessertations about saving 1 nanosecond on some operations

Simd

14:07

@mousetail I didn't know that!

mousetail

You underestimate just how much time and money it takes to really optimize these types of things

Simd

@mousetail I only know about the awesome code on this site :)

mousetail

Try looking up the ridiculus tricks they use to make v8 fast

Simd

Actually I do know a little about that but that is solving a much harder problem

Adám

Hey, does anyone know what the rules are for a valid git user.name?

mousetail

14:11

@Adám IDK, I've used a ton of odd characters and they all seem to work fine

Adám

Hm, then if I ask git log to report the author for multiple files, how would I know where one ends and the next begins?

mousetail

Seems while I can just add a line break to my user name it won't be printed in git log

The_AH

Git username?

Adám

Huh, it appears I cannot ask for multiple files at once anyway.

Simd

@mousetail how far have you got through the implementation

Simd

14:49

actually I can't see how to have a fastest-code factorization challenge for 64-bit numbers as I wouldn't know how to time something very fast

mousetail

@Simd It doesn't work because of a bug in the library's implementation of >>. I'll continue tomorrow, done for today

Basically it crashes I think if you try to shift by more than 64

The Empty String Photographer

@Simd Python’s timeit.timeit runs the code 1 million times. Maybe you could do that as well.

@mousetail The PHD students better do it in 1 nanosecond or it wouldn’t be worth it, since you would waste more time than you have optimised.

mousetail

That makes no sense

Rydwolf Programs

15:04

@Simd FWIW a 64-bit factorization challenge would likely be pretty boring

Maybe some interesting stuff you could do with lookup tables or something, but other than that I don't think it's nearly big enough any of the interesting techniques would come into play

Simd

@RydwolfPrograms why ?

@TheEmptyStringPhotographer good poiny

Rydwolf Programs

^^^

Simd

what do you think of an interesting techniques?

@mousetail ah.. good luck tomorrow!

mousetail

Might need to submit a PR to fix the bug but then it's unlikely to be done by the 1 week deadline

Simd

@mousetail I would definitely report it nonetheless

if you do, please give me the link

Rydwolf Programs

16:24

mmmmmm this is some delicious static typing

The_AH

Delicious not making sense.

B: HttpBody + 'static?

How is that valid syntax???

Rydwolf Programs

A leading apostrophe indicates a lifetime, not a string

The_AH

That's bad syntax.

Not only is it ugly and confusing.

The parser also has to based on context know it's a string or lifetime.

Reduce context as much as possible, kids! - Sun Tzu, The Art Of War.

Rydwolf Programs

No, Rust doesn't have single quote strings

Only character literals

I'm not a huge fan of the leading apostrophe syntax for lifetimes and loop labels, but it's better than a lot of alternatives I could think of

The_AH

BETTER?

Rydwolf Programs

16:28

And once you're used to it it's not that bad

Unrelated String

and speaking of context it's hardly a source of confusion for humans reading either because you literally just won't see them in the same places

The_AH

How about |?

Unrelated String

oh fuck no

The_AH

@UnrelatedString Compiler though!

Unrelated String

pipes as a prefix sound so weird

Rydwolf Programs

16:30

^ and how's that any less ambiguous

Rust's syntax is definitely one of my main complaints with it

Main thing that bothers me is <>

1. Most editors don't do bracket matching for those
2. Hard to type
3. Ugly

My ideal syntax is prefix with spaces, with parentheses allowed for readability

E.g., Map String (Option Box int)

The_AH

Hmm.

Like a scope!

mousetail

What editor can't match <>? All of them can

The_AH

Map String{Option Box int}

That is so much better.

mousetail

Or at least both of them

Unrelated String

it is crazy that rust didn't go in that direction for all the other influence it clearly takes from functional languages

Rydwolf Programs

16:35

@mousetail Lmfao what kind of weird ass assertion is that

Have you personally used every single code editor

mousetail

Both VScode and InteliJ can match them

The_AH

I have used Sublime Text.

mousetail

Even notepad++ and kate can

Unrelated String

like i don't think there's anyone on this damn planet who thinks extraneous brackets for generics and function calls looks better or is easier to use

The_AH

Me.

mousetail

16:36

I've never encountered an editor that could not match <>

Rydwolf Programs

I'm pretty sure VScode doesn't do matching by default for <>

But either way my main editor is CodeMirror based

mousetail

It does

Rydwolf Programs

And it's annoying for regex-based things since <> can also be infix ops

mousetail

Have you even tried running rust in vscode?

Rydwolf Programs

...yes? I use VScode for Rust all the time

mousetail

16:37

Yea it can but it will know it is a generic

So it can match the <>

Rydwolf Programs

Yeah no I just tried booting up VScode and yeah there's no matching or completion for angle brackets

Maybe if you use an extension or something

user

Wait wdym no matching?

Rydwolf Programs

But base VScode doesn't syntax highlight or complete angle brackets

user

Ctrl+F works just fine for angle brackets

Rydwolf Programs

@user Like, underlining the corresponding bracket

user

16:39

Ah

Weird, I'd've expected the Rust extension to add that

Rydwolf Programs

@user Oh it might

Idk

I don't use it since it's massively annoying

And it can't find my WSL Rust install unless I bother connecting to WSL every time I open VS code

user

I think it's fine if the default IDE doesn't match <> as long as your language's extensions do

@RydwolfPrograms I mean, how do you expect it to find WSL Rust unless you open a WSL window?

Rydwolf Programs

Well I don't

user

It's not that painful to open a new WSL window

Rydwolf Programs

But it just means that it's annoying to use it so I don't

→ 20 messages moved to Trash

→ 1 message moved to Trash

Please stop

You're interrupting and cluttering up an on-topic conversation and none of us know what on earth you're saying

@user No but it's just extra friction. And like I said the Rust extension just annoys me anyway

All it does is put errors and warnings everywhere when I'm in the middle of writing stuff

My point tho is that <> is suboptimal as a bracket type for numerous reasons

The_AH

16:44

I do not support brackets ever being used, I will clarify that.

Rydwolf Programs

Angle brackets? Or brackets in general

The_AH

()

Ginger

what

UndoneStudios

why

Ginger

@RydwolfPrograms ?????

what happened there

The_AH

16:48

Automated bot posted a poem.

UndoneStudios

You can post a poem anywhere but here

Rydwolf Programs

Automated bots should not be running in TNB

UndoneStudios

I learnt that the hard way

The_AH

Well posting a few messages does not count as a bot I think but okay I will remember that poems should not be here.

Ginger

...yes it does

our rules are quite clear-cut

The_AH

16:49

Who made that definition! Because...

UndoneStudios

cgcc-se.github.io/chatiquette

The_AH

"software application that is programmed to do certain tasks"

Ginger

if the messages are not sent by you directly through the native chat client it's a bot

Rydwolf Programs

If messages are being posted through your an account by an automated program not directly controlled by you that is a bot

The_AH

Oh.

Ginger

16:49

and we do not allow bots here

it's in the chatiquette and everything

Rydwolf Programs

@Ginger Well a non-native client you still control is fine

mousetail

@RydwolfPrograms I'd install vscode inside WSL, it warns you not to but in practice it works a lot better

The_AH

Runs away in botty panic.

Ginger

@RydwolfPrograms ofc

I just didn't feel like defining "you control"

does anyone else have a feeling of deja vu? :p

UndoneStudios

Does not controlling count as controlling?

Ginger

16:51

no

UndoneStudios

@UndoneStudios (clarifying) this is a joke, do not take this seriously

user

@RydwolfPrograms We might have different setups because I can't even tell the difference when I open VS Code and a WSL window opens up because my last open project was in WSL. I understand not liking the Rust extension though, not all LSP servers are great

Ginger

*cough*Kotlin*cough*

user

Pretty much every language I've used in VS Code lmao, not including Python

mousetail

@RydwolfPrograms Btw the proper extension for rust is "Rust Analyzer" not "Rust" the later hasn't been updated for years and doesn't understand modern rust well. The first is a fork that works properly

Rydwolf Programs

16:57

Ik, I've got that one

mousetail

Ok yea I used the wrong one for way too long before realizing lol

user

The JS and TS extensions actually work quite well

I think they're made by Microsoft, which might explain soemthing?

mousetail

JS and TS are the main languages VScode was made for

You'd expect them to be the best ones

user

Ah

UndoneStudios

Weird

I expected VSCode to be made for Microsoft programming languages (originally)

mousetail

16:59

Like typescript?

user

There's VS for that

Oh yeah TS is literally Microsoft

mousetail

You realize visual studio exists already for C# and F#

UndoneStudios

Yeah

Apparently I should learn what languages Microsoft has made

Anyways gtg now

user

Check out Lean and Koka, they're pretty cool

The_AH

Lean the please.

Koka the coconut.

That makes sense.

The_AH

17:53

The Ninth Byte.

The Ninth Hundred Byte.

The Ninty-Fourth Byte.

Simd

18:13

Anyone tried mojo?

I haven't seen any mojo answers yet

The_AH

Mojo, the Python superset with awesome speed you talkie about.

I haven't because it's too pythony, it's too pythony.

It's too mixy mixy with the see see, see, see.

I want Python but I don't wanna have to get the language to be turned into somethnig it wasn't meant to beeee.

Simd

That's an answer :)

Ginger

Mojo is ridiculous

The_AH

Indeed.

Unrelated String

correction

mojo🔥 is 🤡

Ginger

18:29

🔥🔥🔥🔥💯🤣👌

The Empty String Photographer

😅

New Posts

0

Q: Divisor chain counts

The divisors of a natural number form a poset under the relation of "a divides b?", \$a | b\$. This challenge is to produce the number, \$C\$, of non-empty chains of such posets for natural numbers, \$N\$. This is A253249 in the Online Encyclopedia of Integer Sequences. That may sound complicated...

The_AH

@TheEmptyStringPhotographer Mojo you have been missing out on the dojo which is a bullet for all your children.

The Empty String Photographer

@The_AH I have none…

The_AH

Ha!

hyper-neutrino

18:53

damn the emoji obsession has gotten into language names as well

even emojicode doesn't have emoji in the language name

The_AH

Haha. EMOJI OBSESSION.

🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥🔥

Unrelated String

when is rust rebranding to just a crab emoji

Rydwolf Programs

main.🦀

Ginger

in The Sand Trap, May 2 at 19:36, by Rydwolf Programs

i just ate 🙈 a handful of nails 😺💻

The_AH

Wow.

Simd

19:48

@Ginger what makes it ridiculous?

Ginger

I'm not going to respect a language that uses a fire emoji as its file extension

mousetail

20:08

If you want high performance why not use a high performance language?

Rydwolf Programs

I can't tell if you're arguing for or against Mojo with that statement lol

mousetail

Against I think

noodle man

23:17

can a lang be considered a programming language under this sites rules if programs are day dependent based on real time?

Rydwolf Programs

It would probably follow the same rules as Lost

noodle man

i had an idea for a lang based on the Billboard hot 100 list where each command is one of those

like it wouldn't be that hard to write a primality test but it would only work for the list for that day

Unrelated String

ahaha

noodle man

there migiht be a way to find the list for a given date actually

it's not a fledged out concept just an idea

Unrelated String

also wait does billboard actually update daily

noodle man

23:19

@Ginger inspired by this

@UnrelatedString i actually have no idea lol

but it updates pretty frequently

maybe weekly?

Unrelated String

weekly sounds likely

Rydwolf Programs

> A new chart is compiled and released online to the public by Billboard website on Tuesdays but post-dated to the following Saturday,

Yup, weekly

noodle man

if there's an API, i could just cache the list for that day with an online interpreter ig

Unrelated String

yeah that's also just intuitively plausible because iirc, like, oricon is also weekly

Bubbler

so when a song falls out of traction, all programs using it become invalid forever?

Unrelated String

23:24

presumably the songs’ values as instructions would be pegged to their rankings

so the programs would completely break every week

Rydwolf Programs

^

noodle man

exactly

Bubbler

lol right

noodle man

that's why i was wondering if it would count as a programming language

Rydwolf Programs

So you'd just use BillboardC version 2023-48 or whatever

Bubbler

23:25

now the challenge: write a program that works for multiple weeks

Unrelated String

…also it’s been a while since i looked at oricon and how the hell is idol still #6 on there

noodle man

yeah

Unrelated String

it’s been like 4 months

what the fuck is happening in japan

Rydwolf Programs

Maybe all of Japan is dead and nobody thought to check yet

Unrelated String

plausible

Rydwolf Programs

23:26

I mean when's the last time you made sure all of Japan's not dead? I know it's been ages since I last did

6

lyxal

@UnrelatedString I'd like to make it very clear that what I did on the 9th of October 2023 had nothing to do with whatever's happening in Japan. That was not me.

noodle man

@lyxal No witnesses were found

Transcript for