development

[quartz kindle]

you know whats funny? the geonames api im using requires auth via registered username, and i never made an account with them

im using someone else's username that i found on stackoverflow by chance, and it works

i'll make my own account eventually, but for now im having fun testing gb's of downloads in someone else's name lmao

[pearl trail]

didn't knew tim is evil

[sharp geyser]

He's gone off the rails lately

[prime cliff]

Today i learned that ubuntu 20.04 breaks ps command because it dosen't support date formatting unlike 22/24.
The option is just completely gone and it's not even a standard thing.

[neon leaf]

I just stopped supporting 20

easy fix 👍

[prime cliff]

Was that a mr beast scam.... that's a first

[pearl trail]

https://tenor.com/view/mrbeast-ytpmv-rap-battle-squid-game-squid-game-vs-mrbeast-gif-25491394

[quartz kindle]

my sparql queries are evolving (thanks to chatgpt, even though most of what it says doesnt work)

[neon leaf]

what the fuck is that

[sharp geyser]

so why not postgres?

[neon leaf]

wikipedia

[sharp geyser]

What

[queen needle]

It's for wikipedia's API

[quartz kindle]

exactly

[quartz kindle]

yeah this is how you query the wikipedia api (one of the apis)

[neon leaf]

why not just graphql 🙏🙏

[quartz kindle]

idk ask them

[sharp geyser]

Why in the fuck

Just webscrape at this point

[queen needle]

Scraping wikipedia also sucks

[neon leaf]

prob slower

you could download the wikipedia archives and convert them to a real db ig

[queen needle]

https://www.npmjs.com/package/wtf_wikipedia

[sharp geyser]

tim wanting to make everything himself 🧌

[quartz kindle]

@_@

im just trying to get as much as i need from a single request

so i dont need to do a shit ton of reqs

also, some queries are not possible in other apis

for example queries that use wikidata properties, like get all pages whose property X equals value Y

[sharp geyser]

Made a very shitty stepper component

Not even a component

just slapped it into the page itself

[queen needle]

Can I sees

[sharp geyser]

https://pastes.dev/zqVwBcD4xr

I am telling you it is absolute dogshit

and wouldn't be how the end component would look

but its a proof of concept

It doesn't save state from previous steps either

soooo

[queen needle]

Oh I meant how it looked

[sharp geyser]

It works though

OH sorry

Like I said

Dogshit

😭

[queen needle]

I mean it works it works tho

It's nice

[sharp geyser]

Thankssss

[willow ether]

tsk tsk, using regexes to validate whether something is an email.

[sharp geyser]

tsk tsk tsk I don't care right now

Not to be rude

It's proof of concept

[willow ether]

lol, I was trying to do a funny sorry

[sharp geyser]

Also, it's not really that slow

[willow ether]

But I'd mark it in a real code review

[sharp geyser]

Indeed

[willow ether]

Its not about speed, its about correctness/maintainability.

[sharp geyser]

What other methods are there to check if something is an email? Cause even things like zod and yup use regex under the hood afaik

[willow ether]

parse it character by character using the official spec; its actually not that hard

[neon leaf]

Math.random

[willow ether]

(or just use a library like most people probably do)

[sharp geyser]

I mean yeah most people use zod or yup for that kind of thing

but they also use regex

[willow ether]

I wouldn't use one that used regex though but w/e

[sharp geyser]

Not many don't use regex, its easy and not terribly slow. Also for maintainability I dont think the email spec changes that often

[willow ether]

aaand github went down when I tried to look at the regex they're using lol

I'm actually wrong lol, the html5 spec literally provides a regex for this use case 🤣

<---- suffering from PTSD about regex-based validation mistakes in the past

[queen needle]

I think the best is regex for client side validation and then use email confirmation

So it has to be what could be a valid address and then their actual email/an actual email

[willow ether]

Yeah; my (wrong in this case) nitpick is that if you're writing a validator for something, the first instinct of reaching for a regex is probably the wrong choice for non-trivial tasks.

[sharp geyser]

Normally yes, but for my case no

I don't need to any actual email validation

I simply am doing a pre-check to make sure their email domain is valid before letting them proceed

[queen needle]

I love just using a character by character parser for the dumbest things

[sharp geyser]

Saves both me and them effort & time

[queen needle]

I used one for {user_id} -> 382938 etc

[neon leaf]

I wrote that in pure typescript types smh

type GetPlaceHolders<S extends string> =
  S extends `${string}{${infer W}}${infer RE}`
  ? [W, ...GetPlaceHolders<RE>]
  : []

type Options = GetPlaceHolders<'hello {user}, how are you {day}?'> // 'user' | 'day'

(well close enough)

[queen needle]

How tf

[neon leaf]

its not really complex

its just pattern matching

${string}{${infer W}}${infer RE}
matches *{*}*

the content inside {} gets extracted into type var W

and I just add that to the array, then take RE (rest of string on right) and run the same type again

recursively

this is also useful, turning a record into a string union

[queen needle]

I've never really touched the extreme ts types,chow does that export a string?

Like if I want to send a message using it or something, bc usually js message.send(interpolate("hi I'm {name}"))

[carmine spruce]

y'know the fun part about logging command usages to see what would need attention for optimizations and such is that I can see when the bot got reviewed from the sudden influx of commands that never saw the light of day

[quartz kindle]

thats some advanced ts fuckery

[sharp geyser]

Believe it or not thats considered basic /j

[quartz kindle]

i'll never do that kind of stuff because i use jsdoc

it doesnt support more advanced things :^)

[sharp geyser]

Guys

I am trying to get the user's country, should I just have them supply it or should I get it myself

[quartz kindle]

if its a form, use a select menu

[sharp geyser]

I can get it via a form yeah just wondering what's most convenient

[quartz kindle]

usually a select menu is the standard way of getting someone's country, unless you need to automate that

[sharp geyser]

I don't need to

Right now I am trying to implement stripe

[quartz kindle]

like getting it from their ip address or from browser's locale

[sharp geyser]

and since I don't use hosted dashboards I have to supply the country myself for connect accounts

Tempted to make em all US and commit fraud /j

[quartz kindle]

you need to specify the buyer's country?

[sharp geyser]

The seller

[quartz kindle]

the seller?

[sharp geyser]

Yes

[quartz kindle]

like, the website owner

[sharp geyser]

Account owner

[quartz kindle]

or are you registering sellers

[sharp geyser]

yes

[quartz kindle]

ah i see

then the seller needs to specify it as accurately as they can

[sharp geyser]

I've gone through like 10+ implementation methods and none of them work effectively

Stripe is simply just annoying as fuck to implement

[quartz kindle]

rip

[sharp geyser]

Waiting for someone to either overhaul stripe's api or someone make a better payment gateway cause my god

Solo dev implementing stripe at this scale

[quartz kindle]

i havent messed with payment gateways yet

but i've had my eye on charbee for a while

ever tried it?

[sharp geyser]

Yeah but sadly what I am doing I have to use stripe directly for

There exists no third party that i've seen that does it

Since I am doing seller specific products and not subscriptions

[quartz kindle]

rip

[sharp geyser]

Wanna know what

Fuck it

I really don't need to make the stripe stuff myself

I doubt users will care if they are redirected to stripe

[vale sage]

I can agree with this, watching the progress is soothing to the mind

[queen needle]

Have them to select but autofill what it is based on ip/browser

[carmine spruce]

anyone familiar with incendo's cloud command library?
their server's dead, the docs are.. kinda empty, and I have no idea how suggestions are intended to work

[queen needle]

Send code so users can help

[carmine spruce]

lemme see if I can find the old code I had for it

this is what I had a while back, eventually killed it because it just didn't work at all
it's probably not even how you're intended to do it, but there's also no documentation on how you should

[carmine spruce]

actual code that also doesn't work

the print statement never goes off, so it's not even running it to my knowledge

[carmine spruce]

the lack of documentation on incendo's part is going to be the death of me

either that or the fact it's 2am idk

[carmine spruce]

damn can't get a response from the incendo discord and talking about it here kills the chat 😔

[sharp geyser]

I am starting to branch my backend out to its own thing instead of using nextjs route handlers. Which has led me to use better-auth's jwt plugin. I am wondering how I can manage this in development though. Since the cookies are not shared between the frontend and backend api due to it being on different ports (I guess?) better-auth breaks since the authentication tokens don't exist on the api's context

Which causes 401 errors when the backend makes requests to the frontend api to get a jwt token and then validate it on subsequent requests to protected endpoints

Actually

I am le stupid

The frontend should be sending the backend the jwt which it then uses to validate

[pearl trail]

so, how your login flow works? do you login user to nextjs or your own backend? if it's nextjs, you connect your database to nextjs just for auth?

[prime cliff]

In production are you just proxying the backend like /api -> backend port or something?

[sharp geyser]

I use nginx yes

[sharp geyser]

nextjs handles all the auth and what not.

my backend api is written in C# which I want to protect with the same auth system so i use jwts

Right now any request made to the backend api must be accompanied with a jwt that the backend then validates using jwks endpoints in the nextjs app

[prime cliff]

Then you can just replicate that while developing either local proxy + backend or you ssh proxy the backend port to your computer

[sharp geyser]

That's fair

I learned i dont actually need to care about cookies here

I can fetch the jwks without any auth cookies

SO I can request the json web key set from the backend and validate the token the frontend sends

[wheat mesa]

Auth usually is going to need cookies, unless you don’t care about XSS attacks

It’s the most “secure” way of doing things

[sharp geyser]

Right

What I mean is I don't need cookies for this instance

Next JS requests a resource from backend by sending a JWT along with the request
Backend sees this JWT and validates it against a JWKS
If valid responds with the appropriate resource otherwise responds with a 401

Initially the jwt is gotten from the cookie on nextjs' side

but what I mean there is no need for the cookie to be accessible from the backend since I am just validating the received token

[pearl trail]

through cookies header?

[sharp geyser]

Nextjs will get them from the headers yes

await authClient.getSession({
  fetchOptions: {
    onSuccess: (ctx)=>{
      const jwt = ctx.response.headers.get("set-auth-jwt")
    }
  }
})

When I fetch the session I also grab the jwt on success from the headers

or

I can directly call the /token endpoint with the session token in the auth header

[pearl trail]

interesting. i'd usually do allow cookies to be sent through each request credentials: 'include' and then the server will receive the cookie and get session from db. but every individual have their own way

[sharp geyser]

The thing is I am not rolling my own auth system

[pearl trail]

oh

[sharp geyser]

I am using a 3rd party lib

[pearl trail]

yeah in nextjs right?

[sharp geyser]

Yeah

It handles all the session and shit for me, and yeah I can get it from the database myself if need be, but why do that when I can use JWTs which is a rather common approach

[pearl trail]

yeah i did that, i used that nextjs auth for discord. when an auth is suceeded, i send to server to create a new session with x token, and then set x to cookie

[pearl trail]

if it works, it works™

[sharp geyser]

I mean, my approach isn't much different from others

There's so many ways to handle auth

especially between microservices

JWTs just seems to be the most effective

and its built into the lib I use

Little integration on my part

[pearl trail]

awesome 🔥

[deep cedar]

Hi I'm trying to build an dashboard (my bot is on discord.py) can somebody give me guides,tips, requirements? This would really help me

[compact condor]

get to know html, css and js. Then learn and use fastapi (which is also python and async) as the framework. Use an api instead of websocket.

[quartz kindle]

why fastapi specifically?

there are tons of good frameworks

[compact condor]

sure but its one of the best for python besides django and since he uses d.py I assume he knows python well

[sharp geyser]

Django is also very extensive and not beginner friendly at all

If you've never made an api before

[quartz kindle]

well i dont python so i dont know much about their frameworks, but i hear a lot of names thrown around

from what i heard, a lot of people say good things about sanic and flask

[sharp geyser]

Flask is good yeah

Its more beginner friendly than django

Never used fastapi but i hear its pretty well made as well

[quartz kindle]

and of course, i wouldnt be me if i didnt google benchmarks

xD

[queen needle]

bro could learn python in a lunch break

[sharp geyser]

He could learn it in 10m

[quartz kindle]

lies

[queen needle]

it's very simple, especially with your extensive js knowledge

[compact condor]

flask is very beginner friendly but not solid since every request blocks everything due to flask being sync, for the beginning its pretty good but for moderate traffic it becomes a bottleneck. Fastapi is like flask but faster and fully async supported and uses starlete and pydantic.

[neon leaf]

fixed my missing memory stick 🔥
the issue was a motherboard standoff right below the memory slot causing a short-circuit

[quartz kindle]

lmao almost as much ram as disk

at this point just get rid of the disk altogether and go full ramdisk

[neon leaf]

mm

I could put the 480gb mc server in ram technically

[quartz kindle]

you have a 480gb mc server?

jesus

[neon leaf]

well

I pregenerated 60k radius chunks

literally no purpose though

[quartz kindle]

but... why

lmao

[neon leaf]

idk

I was testing pregen performance

(its basically same as 9900x)

[prime cliff]

128 GB

You have 500 GB+ ram

[quartz kindle]

you know what would be interesting? an online game (even mc) that has its map roughly connected to people's ip addresses

[neon leaf]

what

[quartz kindle]

so that people close by irl would spawn close by ingame

and people far away irl would spawn far away

[neon leaf]

I mean

there are world maps for mc

[quartz kindle]

so you could travel ingame to meet people from other countries ingame

[neon leaf]

in different scales

[quartz kindle]

yeah but have people spawn in the correct place based on their ip address

of course vps could spoof that

but would make for interesting immersive plays

imagine you go on a long voyage ingame, and you arrive in places where all players speak some other language

[prime cliff]

I mean bungeecord lets you connect multiple servers all at once too

[neon leaf]

the 1:500 map is 11x4.4GB parts 💀

well 50gb isnt horrible

there are bigger ones

[quartz kindle]

it doesnt need to be a realistic world map, could be a fantasy one

but the idea is to match people based on their irl location

and make them travel ingame if they want to meet people from other countries

[neon leaf]

what if they just move and stay somewhere else after travelling 5min one time

[quartz kindle]

sure, thats ingame immigration

[neon leaf]

do NOT use bungeecord, its filled with vunerabilities, velocity is the modern option

[quartz kindle]

imagine you emigrate to ingame's equivalent of japan, so everyone around you is literally japanese people

[neon leaf]

I just think this wouldnt work as intended

like people would go where they want

and nothing would truly be regional

[quartz kindle]

they are free to do so, but it would be fun

i have some weird ideas when it comes to game dev

[neon leaf]

yes

[quartz kindle]

another idea i have is artificial ping/glitches

[neon leaf]

???

[quartz kindle]

the further you travel from your ip-based spawn location, the more laggy the game should feel

:^)

[neon leaf]

....

well atleast that would keep people in

[quartz kindle]

and it would make for a heck of an adventure

venturing as far as you can on increasingly glitchy/laggy environment

xD

[neon leaf]

im downloading the 21x4.4gb world map now

[quartz kindle]

xD

[prime cliff]

Finally got Edge Agent to work which means no more open ip/port access.

[lament rock]

Theyre edging

[pearl trail]

huh, do we need to pay to have nested subdomain like x.y.domain.com ?

i'd assume this does not cover nested subdomain?

ah yeah that's right

[quartz kindle]

but it doesnt prevent you from creating them right? it only doesnt cover their SSL

you could still use your own SSL

[pearl trail]

oh yeah, but it's wont be proxied by cloudflare

so the ip will be exposed

[neon leaf]

@quartz kindle u dis

[deft wolf]

Why does this look a bit like a minecraft map? 🤔

[neon leaf]

(it is a 1:500 map of earth in mc)

[deft wolf]

EarthSMP

[neon leaf]

decent ratio

still not very big tho

[neon leaf]

rack done 🔥

[quartz kindle]

xD

how long does it take you to travel the world?

[neon leaf]

it takes ages

if you want you can join

1.21.5 hde-03.infra.rjns.dev:25106

[quartz kindle]

ill need to update my mc stuff

[neon leaf]

mmm

[quartz kindle]

remember when mc used to be < than 100mb

its 500mb+ now

[quartz kindle]

do you have online=true?

[neon leaf]

yes, do you need offline mode?

[quartz kindle]

yup

[neon leaf]

mhm

try now

[harsh nova]

Mmmm yummy

[neon leaf]

@quartz kindle http://176.100.37.105:25129/

[quartz kindle]

lmao

[neon leaf]

what is that render distance 🙏

[quartz kindle]

is was 12 by default, i increased it to max (32) but nothing changed

maybe if i reconnect

[neon leaf]

wait ill increase on server

[pearl trail]

ip yummy

[pearl trail]

use distant horizon with this map, will go crazy 🤪

[neon leaf]

32 render distance

[quartz kindle]

ye i need some graphicsa mods

i g2g take someone to the bus station

[quartz kindle]

ye view distance improved

[quartz kindle]

we just flew from pakistan to brazil in creative mode

took 20 minutes

[neon leaf]

(not worth it)

[quartz kindle]

xDDD

[sharp geyser]

https://tenor.com/bOs6M.gif

[lament rock]

[rustic nova]

is that dynmap?

or something own

[neon leaf]

pl3xmap

[rustic nova]

bluemap is nice

Love you sharex for trying to generate 80mb of clips

but you have like, parallax

street view aah

[neon leaf]

yea

bluemap did around 500cps on this map

which wouldve taken 1 day

pl3xmap did 17000cps

[rustic nova]

topdown only though right?

understandable then

cat enslaving room btw, for teleporting players back after rejoining

[sharp geyser]

        builder.Services.AddAuthentication("Bearer").AddJwtBearer("Bearer", opt =>
        {
            opt.Authority = "http://localhost:3000";
            opt.Audience = "https://localhost:3000";
            opt.TokenValidationParameters = new TokenValidationParameters
            {
                ValidateAudience = false
            };

            opt.Configuration = new OpenIdConnectConfiguration
            {
                JwksUri = "http://localhost:3000/api/auth/jwks",
            };
        });

Was hoping I could just do this

but apparently JwtBearer can't handle the JwksUri thingy

that jwks endpoint returns a json response in the jwks format

[queen needle]

unrelated, but if you're still having issues with stripe I just found this: https://github.com/t3dotgg/stripe-recommendations

[sharp geyser]

Oh yeah I know of it

Basically just keep local cache and stripe synced

Didnt need an entire github repo on it

[queen needle]

ah fair, i just knew you were complaining about it and I didn't know if that helped at all

[sharp geyser]

I mean it’s definitely helpful

It’s just not the current problem I’m facing ahaha

[acoustic bough]

does smb know how to create such a bar chart breakdown with grafana?

I have this rn, but I want different color section thingies depending on what language was used

[neon leaf]

ramdisk 🔥

Run status group 0 (all jobs):
  WRITE: bw=32.9GiB/s (35.3GB/s), 32.9GiB/s-32.9GiB/s (35.3GB/s-35.3GB/s), io=32.0GiB (34.4GB), run=972-972msec

[quartz kindle]

nice

[hidden gorge]

does this look good for our locked access modal

[neon leaf]

zip 🙏🙏

#[inline]
pub fn zip_entry_get_modified_time(
    entry: &zip::read::ZipFile<impl std::io::Read>,
) -> Option<std::time::SystemTime> {
    for field in entry.extra_data_fields() {
        if let zip::extra_fields::ExtraField::ExtendedTimestamp(ext) = field {
            if let Some(mod_time) = ext.mod_time() {
                return Some(
                    std::time::UNIX_EPOCH + std::time::Duration::from_secs(mod_time as u64),
                );
            }
        }
    }

    if let Some(time) = entry.last_modified() {
        if time.is_valid() {
            let chrono_date = chrono::NaiveDate::from_ymd_opt(
                time.year() as i32,
                time.month() as u32,
                time.day() as u32,
            )?;
            let chrono_time = chrono::NaiveTime::from_hms_opt(
                time.hour() as u32,
                time.minute() as u32,
                time.second() as u32,
            )?;

            return Some(
                std::time::UNIX_EPOCH
                    + std::time::Duration::from_secs(
                        chrono_date.and_time(chrono_time).and_utc().timestamp() as u64,
                    ),
            );
        }
    }

    None
}```

[acoustic bough]

put the muted text below the title and bot next to it

[hidden gorge]

i kinda did that on our password locked modal just gotta carry it over

[quartz kindle]

i would put the icon on top and the text under it

or if you want to leave it next to the text, make it a bit smaller and closer to the text

another option is make the icon bigger and put the smaller text together with the title, left aligned under it

[hidden gorge]

i need to find out why the text is so far down

[quartz kindle]

which text?

[hidden gorge]

"Support"

[queen needle]

Vertical alignment messed up

[quartz kindle]

are you using flexbox?

or inline-block?

[hidden gorge]

Uhh

I forget

I just know it uses span

[slow crow]

Hey quick question so I tried using firebase as a database everything is set up correctly the rules on firestone the code everything even put the code into google gemini lol and it even said it’s correct the error i get when i code my bot with firestone is

“Firestone SDK error no persistence storage is available people download “pip install firestone-admin” but it is installed i have done it 3 times lol anyone know any fixes i also uploaded my crash log from firestone to google gemini and it even said it should work so it might be internal

[quartz kindle]

are you trying to use the offline access feature?

[quartz kindle]

check with dev tools? xD

[slow crow]

I’m assuming no but also have no idea what that means im like 6 days deep into discord bot coding

[quartz kindle]

then i assume you dont need it, you probably need to turn it off somewhere

When you initialize Cloud Firestore, you can enable or disable offline persistence:

For Android and Apple platforms, offline persistence is enabled by default. To disable persistence, set the PersistenceEnabled option to false.
For the web, offline persistence is disabled by default. To enable persistence, call the enablePersistence method. Cloud Firestore's cache isn't automatically cleared between sessions. Consequently, if your web app handles sensitive information, make sure to ask the user if they're on a trusted device before enabling persistence.

[slow crow]

Ok my bot is a web app only so then it should be off

give me one second I can send you a screen shot

[slow crow]

Also I meant firebase not firestone lol dont know if you knew what I meant

[hidden gorge]

Not home rn lmao

[slow crow]

this is my firebase rules and the error

When I run pip install firebase-admin it shows I already have it installed

[quartz kindle]

you shouldnt need the admin sdk

[slow crow]

if I remove that then I can't have firebase as my data base correct?

[quartz kindle]

how do you initialize it in your main py file?

[quartz kindle]

you are swallowing the exception and returning your own exception, so it doesnt show what the actual error is

[slow crow]

so i should remove both the line of codes of print

[quartz kindle]

remove the entire try: except:

imports shouldnt really be error-handled, as if they error the app shouldnt even attempt to run

better to let it crash and let it show any import errors

[acoustic bough]

hmm

[pearl trail]

on t.gg

[acoustic bough]

idk how else I should design it

[deft ridge]

why not top.gg

[vestal sun]

I have a custom status on my bot, and somehow it randomly disappeared, anyone else have this before?

[acoustic bough]

your bot reconnected to the gateway so the status was reset

[acoustic bough]

idk maybe

[vestal sun]

Any solution to the status not disappearing like that?

[acoustic bough]

just set it again

[deft ridge]

got a question though, what font are you using for "on" and ".gg"?

[acoustic bough]

https://github.com/Luna-devv/mellow-web/blob/master/app/layout.tsx#L19

[deft ridge]

thanks!

[quartz kindle]

depending on the lib you're using, you can define a "default status" at login

[neon leaf]

@quartz kindle I will maybe generate a 1:1 earth map

depending on if it fits into 4tb

[quartz kindle]

what the f

[neon leaf]

ye

on a sidenote

I have a genius idea for an archive format

streamable and indexable

by splitting it into many smaller indexes that are written every 4mb for example

to get all files, you only read the indexes

though this already sounds cruel to implement

[quartz kindle]

are you sure its "only" 4tb?

that seems quite small actually

[neon leaf]

well

1:500 is 90gb

thats how far i calculated (I didnt)

[neon leaf]

1kb per block is a bit much

especially in newer versions

with incremental chunks

[quartz kindle]

even if its 1000x smaller, its still 510tb

[neon leaf]

well I will see ig

[quartz kindle]

lmao

[neon leaf]

hm

[quartz kindle]

inb4 your computer explodes

[neon leaf]

or I just scan directory before writing data

then I can append an index at the start

in the order the data will be written

[quartz kindle]

brb grabbing food

[deft wolf]

Imagine 2b2t on a 1:1 scale world map

[covert gale]

it will not fit in 4tb

region files are compressed with gzip by default, so you'd wanna switch to something a bit better with space to even attempt that

i.e. zstd

[gilded current]

I have a question

[prime cliff]

You didn't say what the question is

[quartz kindle]

classic topgg

[wheat mesa]

I have many questions

[queen needle]

i too have a question

what was their question

[quartz kindle]

to ask or not to ask, that is the question

[prime cliff]

Tune in next week for the exciting question!

[sharp geyser]

To do schoolwork that is due today or not do school work thats due today, that is the question

[lament rock]

Just do it while it's being collected

[queen needle]

gaslight the teacher

[pearl trail]

tell the teacher that you’re going to submit late and proceed to submit on time

[jaunty mason]

https://wyrd.netlify.app

rate 1/10?

[deft wolf]

1/10

[jaunty mason]

..

opinions

i respect it

[sharp saddle]

good

[jaunty mason]

thanks

i have to update some things

[wispy python]

8/10
The UI looks good, I'd suggest to add more pages.

[jaunty mason]

https://wyrd.netlify.app/#

there isnt much more to add

i updated it

small things

but i dont know what to add

since its just a bot

[wispy python]

take a look at 2nd and 3rd link in my about, you can try making that.

[jaunty mason]

that looks good

i might have to steal some ideas from that 😭

[wispy python]

haha, no issues

[jaunty mason]

https://wyrd.netlify.app/

what abt now

[deep merlin]

I need help in deploying my bot on Heroku. I have command that requires the user to unlock it by voting for the bot on Top.gg. I am using topggpy and discord.py for the bot.

The bot works perfectly when I deploy it on my local machine and expose it using ngrok custom domain. The bot receives Top.gg requests normally.

But on Heroku, when I try to use it's custom domain provided by Heroku, it doesn't work. What can be the reason.

I asked AI and it wants me to make a flask app and receives requests from there, but that still doesn't work.

[pearl trail]

do you set your own port or from env?

[deep merlin]

I have set my own port.

[pearl trail]

try to use the port from env. PORT

might not 100% work but ig worth a try

[deep merlin]

Alright.

Let me try it out.

I will let you know.

[pearl trail]

[jaunty mason]

can anyone be the co dev of my bot 🥀

continuous updates tire me the hell out

[deep merlin]

It worked! Thank you so much!

What can be the reason? Why can't I specify the port on Heroku? 🤔

[pearl trail]

glad it worked. well the reason can be heroku only proxy to the port they pick. if heroku proxying to port A and your webserver is at port B, then it wont work

[eternal osprey]

hey, so i cntinosuly spawning my python process in but its so damn slow to startup and exit the model.

I read that we can keep a python worker open, but to be fair, what are the downsides?

Won't the worker block concurrent requests from multiple users? It's basically going to sequentialize my code..?

[pearl trail]

run py and your program, and communicate? like using grpc or else

py acts as micro service, it runs on its own

[quartz kindle]

if the worker can only process one request at a time, then you need multiple workers and rotate them

[eternal osprey]

can't we aprallelize the requests somehow

[tacit estuary]

I'm just gonna say I hate discord buttons on discord.py and they make no sense on why they need to be split up into 4 entirely different sections.

I literally can only do buttons through chatgpt and how do they get it to show your local time using that <t:1060002345:f> discord time stamp without it looking highlighted in an embed?! I saw that somewhere. I might figure that out eventually.

[quartz kindle]

it depends on what exactly your python script does and how it works

if its cpu-intensive, probably not

if it does a lot of i/o (disk/network waiting) then probably yes

in general, cpu is always blocking, i/o is never blocking

cpu needs parallelization/multiple workers, i/o can live with using async in a single worker

[dusk tangle]

since when we can't use custom emoji in footer?

[prime cliff]

You have never been able to use any kind of formatting in footer lol

[deft wolf]

BotGhost is going through the same thing as Shapes now, only it's about tokens xD

[delicate zephyr]

wait really?

Oh lol

https://update.botghost.com/

[deft wolf]

Yea, I just went shopping, came back an hour later and everyone is talking about it on ddevs

[delicate zephyr]

Funny meme ngl

[tender gale]

just my curiosity but does this mean none of these bots will be accepted here anymore? I mean since now they're offline because of this announcement 🤔

[deft wolf]

We'll see how it goes. BotGhost wants to fight Discord instead of voluntarily changing its system, so it seems to me that soon it won't be possible to create them at all

Most of these bots were not accepted on top.gg anyway because they used "public commands" that were not made by the bot creator

So it won't be a big loss for top.gg (and for the entire bot development community as well)

[tender gale]

good point, thank you for explanation

[prime cliff]

Oh god wtf Shapes had it coming because they did shitty things and message content but now botghost wtf?

I think Discord is overreaching with this token use policy now and it seems like they gave barely any notices or information other than a giant fk u slap it seems.

[deft wolf]

It's possible that these two situations are related and Discord simply doesn't want users to give their bot tokens to third-party services because users are still responsible for what their bot does even if it's not hosted by them or uses someone else's code

Apart from the recent BotGhost drama where people managed to compromise dozens if not hundreds of bots and get their tokens

[delicate zephyr]

the problem is the one point of stress

if all those tokens get breached, depending on scale

the sheer amount of user data that can be retrieved is insane

[prime cliff]

@solemn latch scam

[pearl trail]

wow

[deft wolf]

Likely spammer 😔

[atomic idol]

whos the scammer

[drifting river]

hey im need of hackers and spaamer as friends, lets work and break bread

[quartz kindle]

lmao

[prime cliff]

@harsh nova go bake him a ban

[neon leaf]

zip multithreading 🔥

[neon leaf]

ye

[quartz kindle]

looking pretty 🔥

[neon leaf]

had to use this "sketchy" thingy

use std::{
    fs::File,
    io::{Read, Seek, SeekFrom},
    os::unix::fs::FileExt,
    sync::Arc,
};

#[derive(Clone)]
pub struct MultiReader {
    file: Arc<File>,
    offset: u64,
}

impl MultiReader {
    pub fn new(file: Arc<File>) -> Self {
        MultiReader { file, offset: 0 }
    }
}

impl Read for MultiReader {
    fn read(&mut self, buf: &mut [u8]) -> std::io::Result<usize> {
        let bytes_read = self.file.read_at(buf, self.offset)?;
        self.offset += bytes_read as u64;

        Ok(bytes_read)
    }
}

impl Seek for MultiReader {
    fn seek(&mut self, pos: SeekFrom) -> std::io::Result<u64> {
        self.offset = match pos {
            SeekFrom::Start(offset) => offset,
            SeekFrom::End(offset) => {
                let file_size = self.file.metadata()?.len();
                if offset >= 0 {
                    file_size.saturating_add(offset as u64)
                } else {
                    file_size.saturating_sub((-offset) as u64)
                }
            }
            SeekFrom::Current(offset) => {
                if offset >= 0 {
                    self.offset.saturating_add(offset as u64)
                } else {
                    self.offset.saturating_sub((-offset) as u64)
                }
            }
        };

        Ok(self.offset)
    }
}

because the zip lib only seeked once to the start of the entry

which then obv got overwritten by another thread

so this implements a fake seek that simply stores an offset and uses unix reat_at

[elfin helm]

@harsh nova the profile picture...

?av

[pearl trail]

💀

[hidden gorge]

Omg..

[hidden gorge]

ended up making an endpoint system for a ws like express.js routes

[queen needle]

Oh?

I treating

Interesting*

[hidden gorge]

also each client is assigned an id and a token, to even make a request on that ws you need to supply the ws client token :}

[queen needle]

Makes sense

so how do you create a new websocket route

[hidden gorge]

like this

[queen needle]

ncice

[hidden gorge]

when the WS connects you get this

the server then decrypts it and cross checks it

is this safe/secure though?

[queen needle]

avoid using deprecated methods but I don't see an issue with it, but people like tim and others will know better than I

[hidden gorge]

?

[queen needle]

I was thinking if this was consistent

but since you don't need to decrypt through restarts it's fine

[hidden gorge]

oh well it kinda does now 😭

[queen needle]

if you need to decrypt through restarts like, something is encrypted, restart, needs to be decrypted then you shouldn't generate a random secret and should instead store one in an env

[hidden gorge]

each time a client disconnects and a server restarts the clients token is wiped and reset

[quartz kindle]

are you encrypting websocket messages on the client side?

[hidden gorge]

i am not no.

[jaunty mason]

when ur editing your code editor in itself

[quartz kindle]

what are you encrypting?

[hidden gorge]

the websockets auth token

[quartz kindle]

uh... why?

[hidden gorge]

bc the live chats can contain confidential data that should only be accessed by 2 websockets that need their own tokens

each live chat saves the 2 websocket ids

to even send or receive anything your token needs to be there in each and every single request.

[quartz kindle]

but auth tokens are usually hashed, not encrypted

[hidden gorge]

oh

well im gonna encrypt messages too

[queen needle]

encrypt everything

[hidden gorge]

this is the example of one message being sent to a chat

[queen needle]

Why did you capitalize chatData object keys

[hidden gorge]

honestly, idk im weird like that

i'll fix it

[prime cliff]

Since Discord dosen't like token hosted bot services i'm adding my bot creator to Dev Space now so it can be self-hosted 😄
Time to bring it back.

[queen needle]

I like it

I had a bot creator a while ago

But it used a custom language it wasn't drag and drop, if i were to make one now it would be

[sharp geyser]

This honestly just annoys me I wont lie

I hate scratch

[robust musk]

Hi

[willow ether]

I suppose low code/no code stuff is good for kids, but keeping track of syntax isn't the hard part about programming

[queen needle]

Finishing your projects is

[prime cliff]

Looks at python

[willow ether]

Does anyone but data scientists and children actually use python?

[queen needle]

i used it for robotics 🥺

[quartz kindle]

this

[prime cliff]

Oh look at that you can export it

[feral aspen]

import type { genOgImagePayload } from 'hamoodihajjiri';

export const generateOgImage = (payload: genOgImagePayload) => {
    const hyperLogo = { light: 'hyper-color-logo.svg', dark: 'hyper-color-logo.svg'}

    return ('https://og-image.vercel.app/' + '**' + (payload?.title ?? '') + '**' + '%3Cbr%2F%3E' + (payload?.subTitle ?? '') + '.png?theme=' + (payload.theme ?? 'dark') + '&md=1&fontSize=100px&images=https%3A%2F%2Fassets.vercel.com%2Fimage%2Fupload%2Ffront%2Fassets%2Fdesign%2F' + hyperLogo[payload.theme ?? 'dark']);
};

Anyone knows how to enforce a line break w/ Vercel's OG Image Playground? Apparently, <br/> doesn't work...

[quartz kindle]

have you tried \n?

[feral aspen]

https://og-image.vercel.app/Hello World.png?theme=dark&md=1&fontSize=100px&images=https%3A%2F%2Fassets.vercel.com%2Fimage%2Fupload%2Ffront%2Fassets%2Fdesign%2Fhyper-color-logo.svg

[feral aspen]

https://og-image.vercel.app/Hello%0AWorld.png?theme=dark&md=1&fontSize=100px&images=https%3A%2F%2Fassets.vercel.com%2Fimage%2Fupload%2Ffront%2Fassets%2Fdesign%2Fhyper-color-logo.svg

I added \n w/ %0A; however, it doesn't work.

[quartz kindle]

idk

doesnt look like the title is affected by anything other than basic markdown

[feral aspen]

I think Vercel doesn't support line-breaks entirely

[acoustic bough]

kinda painful how big caches get even when limiting them, default is like 250 messages per channel and I set it to 10 (max 10 mins old), will prob move it to 0 with a keep over limit

[vivid fulcrum]

now imagine the amount of traffic discord has to process in general

if you're concerned about memory usage you can probably roll your own cache implementation, caching only what you really need

[hidden gorge]

done.

[light vine]

DELETE

your token 😭

[hidden gorge]

token is unrelated.

you cant use my system at all.

its locked down.

plus token changes everytime i restart the server.

[light vine]

alr good

[hidden gorge]

aint even a discord bot token LMAO

[light vine]

oh lol

[hidden gorge]

its a websocket request system i made that requries a token

[light vine]

oh i see

[hidden gorge]

finally it works

[sharp geyser]

Watching theos video on just how fucked stripe is, makes it really disheartening to even implement stripe myself 💀

[queen needle]

lmao

[sharp geyser]

I sincerely hope someone at stripe does an overhaul of their "developer friendly" api

[winged sphinx]

Its actually very easy

The huge amount of stuff the API have can feel overwhelming but its just that you dont need all the stuff they offer (most of the time)

[prime cliff]

With 1/3 of it being either "depreciated", having different refrences/names or just weird json overall
Have fun

Make sure to keep signing in to the developer portal every 15 mins and then get randomly captcha'd every 1/5 chance that also breaks on firefox

[sharp geyser]

Very easy? You must have not used it at all then

Because it is anything but easy

There's a list of problems that theo outlines that still hold true to this day

[acoustic bough]

that's just awful with djs..

[winged sphinx]

I am using it for many projects, what are the problems you are facing? Also the Stripe Developers Discord server is pretty useful, they usually respond within 5 minutes

[vivid fulcrum]

yeah I know. node in general is heavy on storing objects. I don't know what the default caching impl does off the top of my head, however it probably does cache data you don't need. the less fields the better

[acoustic bough]

I already try to limit cache sizes a shit ton, but I do have like 100mb of permission overrides per cluster alone

and I'm on bun with JSC rather than node with V8 but yeah still heavy

I use roughly 5gb with 19k guilds rn

[sharp geyser]

What exactly are you using persey

[winged sphinx]

Mostly billing & checkout features

[sharp geyser]

Right so you aren't doing too much, that's not as annoying as its other features

When you have to start keeping track of a multitude of data that stripe is not reliable at giving you

It gets tedious

[winged sphinx]

Like what

[sharp geyser]

checkouts, subs, products, connect accounts, and the webhook events that go along with it that are not sent in sync

You can receive those events out of order, and its up to you to decide if they are even valid events or not

Since not all of them send with "complete" data

or "valid" data since stripe uses temp data in cases where stuff doesn't exist yet and they themselves create

[winged sphinx]

Oh I see what you mean, I agree that having to fetch datas because some webhooks are just empty is annoying but for the webhooks not coming in sync, its often because you failed to acknowledge them so they have to be retried and if Im right you can just use the timestamp to "validate" the data

[sharp geyser]

Not always

Stripe literally just doesn't send them in order of whats happened all the time

[winged sphinx]

Oh really, thats weird I never noticed that with any checkout or subs event

[sharp geyser]

I mean yeah

[winged sphinx]

but anyway I guess you can just ignore them if the timestamp isnt right

[sharp geyser]

There's also the fact stripe will not stop 2 sessions from occuring at the same time

You can create two checkout sessions at once

Its up to you to stop that

[winged sphinx]

I mean, I do not know how you handle it but that sounds normal

[sharp geyser]

No, it shouldn't be normal

You shouldn't be allowed to create two checkout sessions at once for the same person on the same order

Also it wasn't until recently that stripe allowed you to limit subscriptions per person.

So before you had to do annoying checks to see if a sub was still valid, or if they already had one, etc etc

[winged sphinx]

I agree there should be an option to limit it like you can limit the subscriptions but it sounds better to let people have full control over the sessions and so make it up to you to manage the sessions as you wish

[sharp geyser]

What would you realistically need control over?

[winged sphinx]

I dont know but you can imagine for a big e-commerce websites that you have many different sessions and you want to handle which session to restore yourself

[prime cliff]

Usually websites will do a cart session based on the local storage

[winged sphinx]

I see, but both ways can probably make sense

The biggest issue I encountered recently with Stripe is that you CANNOT customize the invoices issued from Sessions based on information that Stripe collect (like customize the invoice based on the VAT/TAX collect)

And it just sounds so dumb when you see all the features you have with invoices and this basic feature is not even available

[sharp geyser]

Not quite sure I follow this reason, there's no way of restoring a stripe session, you have to make a new one if the original one fails.

[prime cliff]

Pog got it working again, had to transfer the 2 year old code and redo the UI

[warped glacier]

So i created a webUI that every second generates me a new unique B64 code. The code is based on a mathmatical equation and multiple object positions. What do yall think?

[queen needle]

What's the purpose?

[warped glacier]

It will be used as a encryption key in the future

[queen needle]

Anyone ever use yt-dlp?

I am using it ```ts
function getAudioUrlFromYTDLP(query: string): Promise<string> {
return new Promise((resolve, reject) => {
const process = spawn("py", [
"-m",
"yt_dlp",
"--quiet",
"--no-warnings",
"--no-playlist",
"--default-search",
"ytsearch1:",
"-f",
"bestaudio",
"--get-url",
query,
]);

let output = "";
let error = "";

process.stdout.on("data", (data) => (output += data.toString()));
process.stderr.on("data", (data) => (error += data.toString()));

process.on("close", (code) => {
  if (code === 0 && output.trim()) {
    resolve(output.trim());
  } else {
    reject(new Error(error || "yt-dlp failed or returned empty output"));
  }
});

});
}

[quartz kindle]

do you want only the url?

its likely downloading the entire video

--skip-download                 Do not download the video but write all
                                related files (Alias: --no-download)

[lament rock]

Also, if this kind of thing is going to be done frequently, it'd probably be beneficial to actually do this in python and expose some kind of api through like HTTP or some kind of IPC since compiling the regular expressions every time for ytdlp is really expensive at scale

[vivid fulcrum]

I'm surprised there isn't a native node port of ytdlp 🤔

[gilded ice]

i have a bunch of possibly-null fields that i want to conditionally assign to fields on this
what's the most efficient way of going about this? i don't think i can use nullish coalescing because the right hand side could be nullish too
i came up with the below but i'm not sure if there's a better way to do this

withConfig(config: IConfig): CommandBuilder {
    this.duration = config.duration ? config.duration : this.duration
    // ...
    return this
}

[wheat mesa]

this.duration ??= config.duration, would this not work?

[gilded ice]

config.duration can be nullish

i don't want to apply it in the case that it is

[wheat mesa]

Ah

[gilded ice]

i suppose... if it's already nullish it probably wouldn't matter actually

hmm

if it breaks i'll be back

[wheat mesa]

I usually do this.duration = config.duration ?? this.duration iirc

[gilded ice]

hmm alright

well

the thing here is that all my fields have default values

i think the above would work

yeah

thank you

lol i haven't used node in years

[wheat mesa]

If config value is provided it’ll override the default, otherwise it’ll stay default

[gilded ice]

yep

looks good to me

[wheat mesa]

Null coalescing operator is quite nice

[queen needle]

So would having a native api written in python be the better solution? I mean it makes sense im just not the brightest

[quartz kindle]

doesnt really matter as yt-dlp is designed as a cli program, not a library

[queen needle]

I just added your line to see how it helps

Dropped times by like a second

[quartz kindle]

welp, run it through some kind of profiler and see whats taking so long lol

[queen needle]

How would I do that? (this is the first time i've ever had to deal with something taking this long)

[quartz kindle]

add "profile" to your options

python -m profile script.py

[queen needle]

const process = spawn("py", [
      "-m",
      "yt_dlp",
      "--quiet",
      "--no-warnings",
      "--no-playlist",
      "--default-search",
      "ytsearch1:",
      "-f",
      "bestaudio",
      "--get-url",
      "--skip-download",
      query,
    ]);
```  Would it go after -m?

[quartz kindle]

actually idk, since it uses some kind of profiler you need to install separately

[queen needle]

Looking at other uses the only thing I can do is deal with it

every reccomendation to make it faster I already have in place, like chunking for streaming

[vivid fulcrum]

just a thought - you're running it through python right? starting up the python environment might take some time - it shouldn't but maybe it does. try running the compiled binary

[queen needle]

not really much change

I tried seeing if I could use invidious but no public api seemed to work

[vivid fulcrum]

toy around with it 🤷‍♂️
I'd try running it myself in a terminal, without the quiet parameter. see the logs and go on from there

it might just be your network

alternatively you could try node packages that do the same thing, I found this https://www.npmjs.com/package/yt-search
seems to be what you're aiming for

https://www.npmjs.com/package/yt-stream or even this. this has parity with ytdlp

[gilded ice]

i feel like i'm going crazy

okay apparently it's not empty

cries

i was going crazy

[vivid fulcrum]

might've been a cosmic ray bit flip 🫢

[feral aspen]

damn~ i haven't seen you for a while

i remember your help back in 2021 :)

[quartz kindle]

web dev is so nice when starting out, but then always turns into an unmaintainable clusterfucking headache

dozens of scripts, components, hooks, routes, backend functions and all that crap for just a handful of actual webpages

[prime cliff]

Depends how well managed the codebase is, what you're using and how you're laying it out

[tropic mantle]

anyone here know how bot page html works?

[quartz kindle]

well, i am first timing svelte, but i've gotten the gist of it

[tropic mantle]

im putting in code but no formatting is showing

can i msg u tim

[quartz kindle]

its just every time i work on web dev, complexity grows way more than expected

lmao

[quartz kindle]

what code are you putting in? you can show here

[tropic mantle]

so yeah whats the problem

[quartz kindle]

get rid of html,head and body tags

[tropic mantle]

kk

is that it

[quartz kindle]

top.gg injects your html into the existing website's html

[tropic mantle]

ah i see

[quartz kindle]

so you need to keep that in mind

you can use dev tools to check the html structure what wraps around your content

and adapt your css rules for that

[tropic mantle]

👍 where can i find that

[quartz kindle]

browser dev tools / inspect element / F12

[tropic mantle]

alr thanks tim

much love

[quartz kindle]

np

[queen needle]

Earlier I had an almost 2000 line css file, it was fun to break up into 4 smaller css files

[quartz kindle]

lmao rip

[queen needle]

Same with a 460 line tsx file i broke up into a bunch of components

https://discord.mx/i/y50a8s5H82.png

33.3 KB (@pancake)

DiscordMX · Jun 26, 2025, 11:55 PM

[quartz kindle]

im trying to wrap my head around reusability, i always think how i can reuse things, so i make them into components and stuff, but then i always run into situations where my existing reusable thing doesnt quite fit the current thing, so i have to either make a new thing from scratch or edit the existing reusable thing to make it fit in both cases, and then edit all existing implementations of thing to account for changes

lmao

[queen needle]

ahh

components with like 200 props so it can be used everywhere

[quartz kindle]

lmao

for example, i have a component for forms

it works nicely for all login/signup/resetpw/etc pages

but now im implementing a user profile page, where they can edit their username/password/profile/etc

and the form component doesnt fit there

because its totally different, both layout and functionality

but both use forms

[queen needle]

i never make form components, i make components used for forms, buttons, inputs, confirm inputs, different types of validation then make the forms seperate using them

[quartz kindle]

yeah i go too deep into reusability and then i get fuged

[queen needle]

i do that too though so i get it

Like i have a renderTrack component that can be used in 2/5ths of the places where I render tracks

[quartz kindle]

yeah i hate when that happens

my brain craves organization, but things always run out of control

then you enter the endless cycle of refactoring xDDD

[prime cliff]

😃

[queen needle]

o h

[prime cliff]

Welcome to component hell
-# It's actually pretty organized

[queen needle]

interesting

https://discord.mx/i/VxlhBm8avn.png

36.22 KB (@pancake)

DiscordMX · Jun 27, 2025, 12:12 AM

my components lmao

don't mind the different naming schemes

[prime cliff]

Why is that single one UpperCase

[queen needle]

listen

I never can figure out how I like to name components

[wheat mesa]

PascalCase for React components, kebab case for Vue components 🙂

Actually no

Maybe you’re right idek how to name shit

[queen needle]

see

[wheat mesa]

Literally been doing web dev for a living for the past year and still don’t know 🙏

[tropic mantle]

anyone have experience with top.gg bot page

im having troubles

[queen needle]

what are you having problems with

[tropic mantle]

thi sis what it looks like:

but this is what it should

[queen needle]

I usually use PascalCase but sometimes snake_case but also like Track.tsx i hate but track.tsx i like

[tropic mantle]

its just not rendering or smth

[wheat mesa]

I use PascalCase for components and camelCase for hooks

[prime cliff]

Maybe the stylesheet or some other request is not loading or loading wrong data?

[tropic mantle]

maybe

can i send u my code

[queen needle]

show us your css your using

[wheat mesa]

Send it here

[tropic mantle]

dont flame me but this is cursor 🔥

[prime cliff]

Is there a link too?

[queen needle]

Fair, idk what the standard is

[tropic mantle]

i dont know how to do frontend

[queen needle]

wait

you said top.gg bot page?

[tropic mantle]

yes

[wheat mesa]

Holy CSS hell

[queen needle]

lmao

[tropic mantle]

yah i dont understand any of it

[queen needle]

i believe you can only use css, and markdown

[tropic mantle]

does anyone have a bot page i could

copy

☠️

[tropic mantle]

does this use smth else

😭

[queen needle]

Javascript and html

[quartz kindle]

i already told you to remove the head/body tags lol

[tropic mantle]

didnt change anything

[tropic mantle]

you cant have css + html?

[quartz kindle]

well, its still one step towards the right direction

[quartz kindle]

you can have css/html, but you cannot have <script>

[tropic mantle]

true i guess

[tropic mantle]

ill try

[queen needle]

ahh i didn't know about html or not

accessibility 100

[quartz kindle]

heres an example of how it should look like

there is no head, no body, no anything else

its only one <style> and then raw html

and all css rules start with .content

[tropic mantle]

alright thank you sm

[quartz kindle]

because .content is the top.gg element where your description is placed

here is how it looks like when rendered in top.gg

[tropic mantle]

ur a goddamn genius

@quartz kindle

[queen needle]

that's what we've all been telling him

[quartz kindle]

@_@

[tropic mantle]

how do you learn stuff like this 😭

[quartz kindle]

experience?

i've been doing shit like this for 10+ years

[queen needle]

jesus bro is old

[quartz kindle]

tell me about it...

im literally twice the age of many people here

[vivid fulcrum]

hey hey! I recognize your name but I don't remember much. 2021 was a rough year 🤣

[vivid fulcrum]

ahh I love c# but god damn is it a nightmare for frontend, at least it was every time I tried to use it for frontend

[deft wolf]

You are not 48 right?

[feral aspen]

It was ;) - I hope you’re doing well.

[quartz kindle]

lmao, not yet

[gilded ice]

hey all

me again

is there a way to avoid exporting a single method from a class?

export class Shocker {
    // ...
    static NewInternal(username: string, apiKey: string, id: number) {
        let shocker = new Shocker(username, apiKey, id)
        
        let url = new URL(WS_URL_BASE)
        url.searchParams.append("Username", username)
        url.searchParams.append("ApiKey", apiKey)

        shocker.brokerUrl = url
        shocker.socket = null

        return shocker
    }
    // ...
}```i've tried doing `module.exports.Shocker.NewInternal = null` but surprisingly mixing cjs and esm like this doesn't work i wonder why

it sounds like an xy problem because it probably is one :clueless:

[pearl trail]

without static, you can do #FnName(...) and call inside the class using this.#FnName()

not sure with static

[gilded ice]

doesn't seem to work unfortunately

it's not the end of the world but it would be nice to have

[pearl trail]

oh my bad im wrong

[gilded ice]

hehe oops

[quartz kindle]

wdym avoid exporting a single method?

[gilded ice]

private but within the given file instead of just within the class

[quartz kindle]

you want to export the class but not the static method?

or what exactly

[gilded ice]

yeah

i figured out a better way of getting what i wanted

[quartz kindle]

the only way to do that is to put it outside the class

[gilded ice]

yep

[pearl trail]

revolutionare

[gilded ice]

i'm so smart

[quartz kindle]

i broke my vsc

jk, its just a binary file

[neon leaf]

I fixed my vsc

(variables are finally properly highlighted in format!)

[gilded ice]

this seems normal

sincerely, a nvim user

[quartz kindle]

huehuehue

[gilded ice]

i wrote my first node library in ages

it rocks

i love using rust design patterns where they weren't supposed to be

i love that node doesn't have a borrow checker

[earnest phoenix]

yo guys, is there somebody here who is looking for/building complex economy bot

[gilded ice]

i'm sure there are hundreds

[earnest phoenix]

oh but i do really mean complex

[neon leaf]

I mean, essentially everything is just wrapped in an Arc

[gilded ice]

i'm sure there are tens

https://tenor.com/view/arrested-development-david-cross-tobias-funke-there-are-dozens-of-us-dozens-gif-5426075

[gilded ice]

but it's even more flexible than that

[earnest phoenix]

how many of them are on chain on their own blockchain tho 🤓

[gilded ice]

i'm sure there's at least one

[neon leaf]

if u want the same then its Rc<RefCell<

mutable shared variables

on a single thread

[gilded ice]

kiiiiiiinda

[wheat mesa]

Node doesn’t have a borrow checker but it also allows you to shoot yourself in the foot

[quartz kindle]

i love having the power to do that

[lyric mountain]

hey tim, did u know, sqlite doesn't have types

[quartz kindle]

yes, they have "hints"

perfect for js

[lyric mountain]

no better duo

[hidden gorge]

Hey.. so we failed to account for almost everything and all our systems collapsed in on each other and we gotta restart, we made the mistake of coding other systems before vital systems. What should we start with?

[quartz kindle]

more context?

[hidden gorge]

Our website, API, Livechat System, Everything broke

we dont know where to restart and what to focus on...

[quartz kindle]

broke how?

[hidden gorge]

failing to work with each system, can barely intergrate with other systems, code is hard to update without errors leading to downtime on production server

and the fact the each system has its own database (which im seeing now was a very bad idea)

[quartz kindle]

well, a lot of things could be learned from that, without knowing more details its hard to point them out, but you should be able to see many areas in which you can improve and do different now

one thing that comes to mind is that exposed interfaces should rarely ever change

for example if you have two systems that talk to each other via messages, even if you need to change them completely inside, the messaging interface should never change, if possible

so that changes in one system dont require changes in other systems

having multiple databases is not an issue as long as you have cascading effects implemented

ie if db2 data depends on db1 data existing, on db1 delete immediately delete from db2 as well

[hidden gorge]

8 databases...

only a tiny fraction

[quartz kindle]

if they are all hosted in the same system, consider using tables instead

[hidden gorge]

i think im gonna switch over to a central database

what sucks is i've been working on all of this since june of last year... :{

[quartz kindle]

welcome to refactor hell

[hidden gorge]

what should be mainly focused on first, we ended up creating the article api before the login api should we focus on user systems first before others?

[gilded ice]

elixir type guards in js 💪

[quartz kindle]

that depends

if you want to launch as soon as possible, create the content first and launch it

then slowly implement automation

[hidden gorge]

we wanted to launch last month 😭

[quartz kindle]

rip

[hidden gorge]

but my bloody COO/CTO just stopped doing shit...

[quartz kindle]

rip²

[gilded ice]

lol

[neon leaf]

10th grade graduated

[pearl trail]

wow you're in high school?

[grand mango]

Me wait 2 weeks for my api

So i run every command smoothly

• antinuke

And whitelist stuff lol

[neon leaf]

idk, German system works differently

[pearl trail]

ohh i see

[prime cliff]

8 databases..... wtf....

[pearl trail]

i wonder will that much databases is actually better than only 1

[prime cliff]

On the brightside even major devs cough hytale fail in all kinds of epic ways with all the resources and money they need.
Having 8 databases though in 1 big system is shockingly amazing and cursed

[hidden gorge]

Each database also stores data differently 😭

so i gotta make a set standard

[prime cliff]

No way who could have guessed that

[wheat mesa]

don't vibe code

biggest advice I can give for making large systems

AI is alright to use every now and then for things that don't matter that much, but don't let it dictate large portions of your code

If you need reliability, LLMs don't provide it

I can see a lot of places in the app I work on for my job where AI was used and it created less-than-optimal solutions

[quartz kindle]

can you share the list of databases you use and what you use them for?

maybe there are things you dont need in there, or things that can be combined

[wheat mesa]

it might also be worth mentioning that using nosql databases can be tricky

[hidden gorge]

I don't vibe code... why do you keep saying that I am???

[wheat mesa]

From what you've posted in here it looks AI generated

[hidden gorge]

waffle..

don't make assumptions without proof.

[prime cliff]

he only posted AI gen coded web stuff i believe not database?

Although tbh doing it that way does indirectly lead to multiple databases lol

[hidden gorge]

I only created separate databases because at first I didn't wanna store account data in the main db for articles

and other stuff

livechats even have their own db for that reason

but i realized what i can do now to be able to store them all in one

[wheat mesa]

I mean I'm assuming that this also tends to mean that AI is used for backend as well 😉

[hidden gorge]

I will admit in the past I used AI 99% of the time but recently have been moving away from it as it lead to nowhere and broken code....

[wheat mesa]

And for what it's worth, I'm not trying to bash you for it, I'm just trying to give you advice for improvement

I had a phase where I was using AI quite a bit

Then I realized how terrible it was after a little while

[queen needle]

Real

[hidden gorge]

I will be honest again... a tiny bit of our CSS is AI

[queen needle]

I genuinely didn't believe I still knew how to code until my AP CSA exam, and the practice for it because I had gotten so bad with it

[wheat mesa]

I mean I've seen some screenshots of stuff you've posted that I can tell was AI generated or at least partially

It's not a personal attack, I'm just saying to use it with caution

[hidden gorge]

I didn't wanna say it because I felt like i was gonna get judged....

[queen needle]

I think it's good for helping you optimize certain things, help with implementation of an algorithm, or vibe coding a simple like one maybe 2 page site you need

[wheat mesa]

I think that it's okay-ish at solving niche issues that are difficult to describe to google

Or at least being used as a "rubber duck"

[queen needle]

It helped me get the implementation for the algorithm to check it a 15 puzzle scramble is valid because I kept reading the algorithm description and was confused, but then after I understood it from the code it gave I created a more optimized solution but I don't think I couldn't have gotten there without it making the first version

[hidden gorge]

I spent 3 hours trying to fix a really dumb bug so I just asked it to fix the bug bc I was fed up at that point..

it also kinda explained some more things to me and also why im now on ESM :}

[wheat mesa]

The primary issue with AI imo is that it just turns off your brain

[hidden gorge]

it made me stupid...

[queen needle]

Real

[hidden gorge]

almost as stupid as webster running rm -rf on his vps...

[wheat mesa]

Programming is 85% critical thinking skills

[queen needle]

But the algorithm for 15 puzzle is stupid

https://www.geeksforgeeks.org/check-instance-15-puzzle-solvable/

[wheat mesa]

There are times where it helps, but I do think that the biggest issue is that it encourages you to offload your thoughts to a machine

And at the very core of a good software engineer is the ability to work through issues by thinking about them

[queen needle]

I tried so hard that algorithm was stupid

I understand it now, and could implement it myself but before I couldn't

I also like when I try to get AI to do something and get fed up with it taking so long I just do it myself

[hidden gorge]

1. Support Database - Just stored tickets and agent accounts as well as Support System API Keys,
2. Website Database - Stored User accounts and Articles
3. Development/Testing Database - Used this during the API Rewrite
4. Livechat Database - Stored Livechats
5. Survey Site Database - Stored out Survey's and responses (WILL BE STILL USING)

[wheat mesa]

Yeah this really shouldn't be separate databases

These should be separate tables within one db

[hidden gorge]

im leaving the survey one seperate bc the CAO is in charge of the survey stuff

the rest im doing in a main DB

[wheat mesa]

That's just overly complex for no reason, you don't need a separate database to work concurrently

[hidden gorge]

(This system is also in production and is currently storing responses needed for reports)

[queen needle]

Do you have Prisma experience?

[hidden gorge]

No.

[wheat mesa]

Separate databases are more warranted when you have services that don't need data from multiple dbs, and are completely independent of one another

[hidden gorge]

well the survey db aint even interacting with the main site 💀

its own its own site

on*

[wheat mesa]

Still, if it's tangentially related, you probably don't want a separate db

[queen needle]

I've always just used different tables

It's easier than multiple databases

[wheat mesa]

For example, what if you want an admin panel on your main site that compiles data from your surveys into reports? You'd need something that can bridge the gap

Either that or now you have two different connections open to two different databases that could've been one