#development
1 messages ยท Page 306 of 1
Yuh, I hopefully have some people ahah
Audit log page
Still needs some work but it works for now
this seems new
I havenโt really paid attention
How's this lookin chat
I'm trying to make it as transparent as possible
Like I don't want a seller to wonder "Hey, it says 693.xx why am I only getting 600"
clean but why so many boxes
Honestly thought it would look better that way but do you think I should make it one card?
veld's right
btw as someone who's been in shit industries @sharp geyser , make sure to protect yourself so your platform isn't liable.
Ima be so real
this is where I am struggling
Making a TOS, Privacy Policies, and all the other policies related to financial means
I have 0 clue on what's good / good enough
where are you based?
United States
imo either a company or DBA so you can protect YOURSELF
but maybe there are some experienced people who've done more
Well I am already incorprating as an LLC
hm
Yeah
I jsut need to come up with TOS, Privacy Policy, Refund Policy and any other policy that's required to basically at least "cover" my ass
LLC is good
I don't wanna get screwed in fees and shit ๐
It's basically a sole proprietorship but with protections for personal assets
I mean that's the risk of creating something
It is good as a one-person company
Yeah
I am forming an LLC under Mercatus LLC or Mercatus GG LLC whichever one the state accepts
Hopefully its the first one ๐
wasn't there a website
or rather a company that do everything for you?
you pay a small fee
can't you check for name before registering ๐ ?
Yeah fair enough
and I would not put it past them to deny it solely because they want too
I went LLC
I am also funding this project out of pocket
sooo
No seed funding yet
as I said for you probbly smarter
I jsut need to find out where I can hire a lawyer to help me with this shit cause I don't want to leave myself areas to be screwed.
Especially when dealing with money ๐
and user data
how long does it take to review and make the bot public on there site?
Dang I thought I could use the tags command
oh well
Me get experienced role when /j
i think you either have to no life this server or when the mods feel like giving you it
actually a horrible spec
spent 30min until i actually understood wtf its sending and wtf it wants
these descriptions really aint it
no remove the box completely from everything but the payment breakdown
gotcha will check in a moment!
AI has made programming boring
Yeah
I see what you mean
keep the padding
Definitely looks a lot better
horizontal padding
on the very top one?
Wait
Like this?
alr yeah this looks better
I think the massive Confirm Handoff button looks weird right?
Gotcha
Yo guys i went for the most minimalist design and reduced the site size to just 35kb
What do you guys think?
also one question should i verify it before or after adding it to top.gg
interesting link
Y ๐ญ
sorry not trying to be mean, i just rarely see .cc around so i might be more vigilant when seeing one
Oh i kinda like .cc
Please donโt make me add another domain to my huge collection
๐ญ
xDD sorry, that's totally fine
as you should ๐ you should always be skeptical of links people post here!
Especially if it comes from me
if i wanna be super safe, i just open an incognito tab on my browser and paste it there
it doesnt matter if it comes from you, my friends, or anybody else, people's accounts can be compromised
i don't think the cookie banner is necessary for just normal uses
I added it mostly for my payment gateway verification
They check through the site
I dont want them to miss that part
I canโt use stripe so i gotta go with lemonsqueezy with much higer fees
y'all like playing snake? ๐
https://sh.veerack.dev
-# open the snake app and start a AI Battle game, lmk how far you can get!
vibecoded asf
how can you tell so fast? 
you can't really just say "you can tell from the design" without getting into specifics ๐
ai has very specific design patterns
like what?
a grid box is a specific AI design? i don't get it
yes AI is involved, if you look at the portfolio main page i clearly specify how i use AI anyway
it always uses a blue/gray gradient (and always tries to use a gradient for some reason), i dont know how to describe it but theres signs on sites that use AI
it tries to mimick shadcn a lot
i love that you're being transparent 
yeah idc, i use AI daily for my projects, but i dont just hey do this for me, i actively debug and design as well
that page design for example was made in figma
you're right, AI added that gradient when it reproduced the design in the webpage for whatever reason. i ended up liking it
i don't want to cause an argument or anything, just wanted to point out that AI has these specifics that become really easy to notice the more you use it
unless you feed it exact specs of what you want, it will default to gradients and blue/gray (even if you specify dark mode etc) and these components that are essentially trained off of the most popular components (shadcn rn)
yeah i know, i've been using AI for the past year or so, i can notice as well but sometimes that gradient effect fits for whatever reason lol
at least its not purple
the UI there is still subject to change anyway, i spent an unreasonable amount of time doing the math for snake and other things there
this is the main vibecoded colour lol
purple and blue gradients are AIs absolute FAV for some reason because its deemed 'modern'
yeah no i can relate
"create me a website where i can list discord bots"
that's the only prompt you need to get that
something that's a dead giveaway for a vibe coded site is emojis
chatgpt loves emojis it pisses me off
Made a funny little mod bot that tracks when users send the same message multiple times in a short timeframe and times them out for a week because that's usually the behaviour of scam bots posting like those crypto scams or whatever.
Also detects images using a perceptual hashing algo and a leven distance on that hash to determine similarity. So far, it's gotten an innumerable amount of scam bots as the guild it's mainly active in usually sees like 3 scammers per day trying the exact same thing. And then only like once a week does someone get past.
I know I could have made a honeypot channel, but people are stupid. And outright banning isn't a good solution since a week is when people usually get their account back. If we notice repeat actions, then we can ban. Mod pings from users being so low now is really nice.
url is localhost because I forgot to update it to the public facing url, but it works out since that's on my home network :)
emojis are the only way ai can express its feelings to you, dont shit on them for that
reddit based comment
thats my best for the record anyway
my best for arcade atm is 80
the game is good, though sometimes when you press too fast it doesn't register your inputs
deploying some changes i had locally rn for that, been making some changes to it for the past few days
should feel and look better
i think the badge and score give it away lol
Insane work, I respect it
love it :D, similar thing I've built for my rules thing in the dashboard, just + deleting ghe message.
more gradients please, I could almost read the text
there is something going on with discord... i am getting these randomly. Seems like random join events are being triggered?
lol
that looks quite serious (relatively speaking) chances are its probably something up with your bot, log the raw events coming into your bot
Those arenโt my bots
Those are 2 different bots
Sent me that message at the same timeโฆ
The bots arenโt related to eachother either
I know the owner of appy
This is surely on discordโs end
If it was from discord's end it must have caused many other bots to dm everyone. So it's not from the discord.
2 seperate bots welcoming me at the same time for 2 different servers doesnโt seem very local to me
It could be a very specific action from me that triggered it
Exactly!
Okay boys i am moving away from frontend. Trying to solely become a backend developer.
What do yall suggest? I got python, ts already on mytoolbelt.
Like a language that will not make me homeless in the future ๐
Frontend gave me a headache. Space is moving too much so i need some more stability
C# is a great backend language especially for web development
It never hurts to know C++, though to get a job in it you have to know it pretty well which can take some time
Java is also a must-have for web dev
+1 on C# for the backend
lmfao probably your code
it aint happing to me
wdym
I know but people can fuck up on their code and not realizing it and no one has reported the bug
Its not Discord end its Developer
i had same problem until i added a return
really easy fix but most likely if the bug isnt gonna get reported to the developers its not gonna fix itself
I mean if it were Discord's fault then it would be reported on ddevs but I can't find anything reported like that today
its not discord fault
its bot devs fault
cuz the bot doesnt know when to stop until a return is added or something like that
its not discord it's the developers of the bot

Hmmm i see. Are there lots of jobs available for it though?
Been hearing lot about rust too
You'll have more chances in C# than Rust
C# has a ton of enterprise level jobs available
yeah unless you know where to look rust is still very new to the job market
most companies ride microsoft so the natural language for any programming ends up being c# for many companies
I see thanky iu guys!
Always there to help me much love
I will start with expansding on python first
i've noticed this our end too
not sure why it happens
Oh wait, member onboarding @slender wagon maybe?
why would it do it for 2 different servers?
hm
What should I add, of remove
The text seems too stretched
Ohh I loved that 
Iโll try it out
its bold branding, if it fits ur vibe its cool
Hi
yo
ai generated yah?
like i can sense it ๐
No

too squished ui imo
add an arm and remove a leg
Decided to build my own reporting system for discord as we run a lil community of servers
Hoping to build out a dashboard with some cool metrics which i find interesting as I'm a nerd.
removed it :) thanks
well yea its in the same vibe as everything
blurs the name but its fully readable
all that work 
it took me so long to blurr it
i just left a part open so you can see the font
He meant here
not only out of place, but a very dangerous statement
*free forever, until i decide to not to
alr
cooked idk what i did there
How do u guys like my minimalistic design
nice blur and 'free' promotion
U must be the funnies person at a party!
Ugly design ngl
I tried keeping it under 35kb
For what reason?
Speed
So it could load on an apple pen?
Yes
i donโt think a few kb matter that much
They do
Especially with the trade off being not attracting to customers
Those kbs will save you loading speed but deter clients away
Not necessarily
I'm fine with minimalist
Alot of people love minimalism
I'm a bit tired of every website being some flashy mix of gradients and overused animations
Or uneeded effects/animations
Seems like every product I see now doesn't even tell you what it does on the home page, you can't tell what half of people are even advertising on their own site
!!
Its cuz most devs want to impress people
But forget why they are making the site
AI tends to put gradients on everything because it's easy to impress people
I appreciate a good low-detail & fast site
U forgot emojis
Taking a look at your site it's pretty good. Only complaint I'd have is that I don't know the source of your metrics of things like "99.9% detection accuracy" and not sure if I trust the claims of "No personal data is ever stored" given the nature of your bot
Yes i was going to correct these.
All the important details about a user are one-way hashed
In the privacy page i mention a bit about it
Not sure if I'd believe that either though, since I'd be wondering how your premium "live feed" works for a given user, since it mentions their username
Usernames arenโt
One sec
I wouldn't say it's bad to retain personal data in this case, since I'm assuming you have to maintain some sort of blacklist or track user activity in order to identify it as risky
Im not on my pc rn but i showd @acoustic bough examples of ehat i store
I compare hashes
hi
We are talking about my encryption
I guess it mostly depends on how you're scoring users' risk
In my mind, keeping track of suspected alt accounts would require storing some degree of personal data
But I don't think there's anything wrong with that in this case, just that the claim that no personal data is ever stored I'd be suspect of
Im scared of db breaches so i said to myself why carry the risk of being fined a good amount of money by EU
Itโs relieving hashing the data
Ah okay after reading your privacy policy I understand more now
And i envy apps like signal
Interesting bot concept, I like it. Are you using any sort of ML model to "score" users, or is it some formula you've come up with on your own?
Iโll show u
Iโll send it in dms
๐
With a salt I suppose?
you'll love geocities
lets normalize that design language again
just some feedback (sorry i didnt get around to your dm) hashing the ip (assuming its ipv4) with sha256 can be easily reversed even if you use a salt because
if you exclude private ranges you only get around 4 billion combinations, and since sha256 is very fast youll crack that in no time
youd be better off using a pepper for that (some kind of secret you store securely that isnt stored in the database)
but honestly me personally i consider ips to be basically public so storing them plaintext isnt usually the end of the world, but totally fair if you want to secure them in some way from data breaches
and if youre storing unique identifiers that are sufficiently long and random (referring to the fingerprint/id hash) then you dont need a salt at all!! but i wouldnt touch/change what you are doing already
Is it good dashboard or just too complex
that looks so complex
make an acutal dashboard
Well it will need a website login for dashboard, i wanted users to manage from discord
like no need to login with accounts to webs..
and it results in something unusable and overly complicated
hmm so you think users would like a website dashboard more
from a user perspective, i see this control panel - i just kick the bot and get another one that has a much simpler way to change those settings via e.g. a web ui
you're free not to do it, just make it less bulky
there's just way too much being shown
https://www.youtube.com/watch?v=C7gJ_UxVnSk can some of u smart guys fact check this
Star the repo on GitHub!
https://github.com/clockworklabs/SpacetimeDB
My Twitter: https://x.com/TylerFCloutier
Our Twitter: https://x.com/spacetime_db
Our Discord: https://discord.gg/spacetimedb
Website: https://spacetimedb.com
Referral Program: https://spacetimedb.com/space-race
Keynote Benchmark: https://github.com/clockworklabs/SpacetimeDB/...
youtube recommended this to me
their benchmar essentially compares spacetimedb vs drizzle
not against any db directly
Epic 
tf is drizzle
such bold claims are almost always not true or with huge caveats in databases
only so much you can do to speed up IO and b trees
an orm
also, their benchmark is network based
so they are actually benchmarking the network stack, not the db itself
Better now i guess
man it's too much button
can you split it into sections w/ dropdowns?
or do it in a modal with the new select components
Menu better?
I think choosing between buttons more clear
ahhh thats too much hochpoch
i would highly recommend moving all buttons into diff menus
that's so convoluted
i love orms but hate them because they can be incredibly inefficient
i hate them regardless of efficiency
:^)
efcore in c# is especially atrocious, the overhead it adds horrifies me
but at least you get a good developer experience
drizzle is fine, its quite literally just type-safe sql
no weird on-app joins or whatever
In order to implement my bot you have to.
Manually create the verified role
Manually create the verification channel where the main embed will be sent
Manually remove the permission for the everyone role to see all the channels & categories
And ofc make it so the verified role can access the rest of the channnels
AND on top of that you have to drag the bot role on top of the verified role
I feel this is wayyy too much for the average attentionspan of a gen z
I was thinking of making a semi-automated setup but idk how to properly implement it.
Do you guys have any idea
I do have a dashboard as well btw.
the bot should be able to automatically create the role, the channel and remove the permissions, you just need to give it the proper perms to do that
a /setup command could let the user select between manual and automated setup
and a /status could list the requirements and their statuses/availability
its a lot of work but its totally doable
i did a similar thing in an old captcha bot i made
ive automated all that with a single command lol though i do have a massive db and alot of fields auto populate lol
Yep ima do that
Is there a way to get the role higher up or does that require manual work
This is one of those things that the user has to do manually afaik, otherwise it would be abused
yeah you need to set the bot to require manage roles, so when invited it will autoatically get it, and then an admin needs to place it above the roles it should manage
or the bot itself can create roles below it, but the admin still needs to place them above other relevant roles for them to take effect
Okay sounds good
Hey everyone โค๏ธ
Quick fun question from me - I saw a few bot pages around that managed to somehow update the look and the icons used on the standard invite and vote buttons on their bot pages.
I only found out recently you could supposedly target them and change them via the long description inline HTML/CSS, but I can't find IDs/classnames to do that specifically, and was wondering if anyone has any advice ?
Thank youu !
Ctrl+Shift+I (Dev Tools) and inspect the elements :)
Then it's just your creativity with CSS, note that some obvious things are not allowed, e.g. lots of flashing effects and such
I did exactly that but I'm still not sure how it works
For example the Invite button looks like this
<div class="w-full md:w-auto css-0">
<a class="!button !button-sm md:!button-lg w-full md:!w-auto button-primary! truncate" href="/bot/1332364544869142538/invite">
<h2 class="hidden md:block">Invite</h2>
<h2 class="md:hidden">Invite</h2>
</a>
</div>
But these don't have IDs and they seem mostly generic classnames that might be used in other places ๐ค
For the example above how would I go about adding a custom icon/src ?
Sorry if it's a stupid question I'm genuinely amazed how people figured out you can do that lol

kurwa?
you can use any combination of ids, classes, tags and props in css
the more specific you are, the better
for example for that specific block you could do:
div.w-full.css-0 > a.truncate > h2 {
...
}
of course you would need to match it from the beginning, so add its parent, and parents parent, etc... until you have a truly unique sequence that can uniquely identify a specific element
you can even do this:
a[href="/bot/1332364544869142538/invite"] {
}
its just standard css
what topgg does is it takes your long description, and places it inside the existing page html
so the page looks like this:
... existing topgg html
<div> some topgg div
{longdescription}
</div>
... more topgg html
so whatever you write in it, it will be parsed like real html once its placed inside the page
and this is 100% valid native html+css ```html
<div>
<style>
a[href="/bot/1332364544869142538/invite"] {
color: black;
}
</style>
<p>my bot</p>
</div>
Small tip, right click the element and you can copy the CSS path/selector for that element
Huh?
He need to customise the vote button on his website
Iโm asking him to create website using html and css and if he need vote counts he can get it from api
No, asking about customizing
"For the example above how would go about adding a custom icon/src ?"
"Huh?" wasn't for you
ah lovely deleted messages
I should start using reply function so it would be more obvious 
hey, hypothetically speaking,
suppose a wordpress site was hacked. I login, i just check the dashboard etc, is there any chance that any malware is injected into my mac? Using safari browser.
May be a dumb ass question
but i am genuinely interested to know
I know js injetion is a big thing, but aren't browsers hella sandboxed now?
my general knowledge: you cant get a virus by just visiting a page. it can download stuff for you, sure, but it cant open it or do anymore
yeah, mac has a strong: "do you want to install ..." ruling though, so i don't think that will go unnoticed.
it can do popups and stuff though, which is annoying but nothing bad
i mean js injection.
mac is pretty secure anyways so its not a bad thing
but even then browsers are sandboxed. Iiirc all tabs are contained in their onw environment
safari has each tab in an environment so yes
unless theres a zero day exploit its impossible
yeah i figured so
i am currently writing a thesis about js injection, skimming specifically
that shit made me paranoid
hai waffle!
how's your engine?
if you want to pentest your own browser there is a project called beefproject on github you may check it out
holy
hmm
that seems legit website
seems legit to me too xD
its on their github linked i just checked
/beefproject/beef on gihub
want help setting it up?
but your gpu spike from just visiting that site idk
well no wonder
its got a fucking full screen canvas animation
with true resolution
holy ๐ ๐
don't they test it though
i bet most visitors have their browser run on integrated gpu
๐
this thing probably uses more power than crypto mining scripts
whats your screen resolution
1440
and your browser is full screen?
im on a minipc
idk, maybe its throttled like crazy
what is your integrated
radeon? its 6969 times better than intel
i7-13700T

i got this for free that cant be beat
the canvas scales to browser width, if the browser is windowed, it wont be as high res
thats a beast igpu
i have to use gamescope and run games on 720p to ba able to play
heavy games
cs2 and farlight84 i have been able to run so far
im on an i7-12700H but i only really play league and maplestory
my cpu is a laptop cpu you can say
https://www.intel.com/content/www/us/en/products/compare.html?productIds=230492,232101 yeah i guess yours is faster than mine
mine literally is a laptop lol
No wonder the stink permeating off of you was intense, you play league
indeed
How dirty is your keyboard
eh, some dust here and there
heh why did my antivirus mark ngrok as a scam?
it wasn't ngrok, a file installed through ngrok.
this was the file it decoded to.
ngrok is flagged as PUA because it's used a lot for its tunnelling feature maliciously
It marked it as PUA, not scam or virus or something similar
i see
i ddo think that this file is the real deal though not a scam or something
bin.equinox.io is the official ngrok download path.
probably installed it when i used brew install ngrok
yeh
microsoft is worse
20 million different types of auth
all in weird submenus
with multiple api versions that arent documented
nah itโs easiest
probably the most basic oauth implementation
But hard if you never used it lol
Yeah though the documentation is pretty well written to be fair
You realize any lang at this point has a library to use discord oauth so saying it's hard means you really havent googled xD
thats 1 of the 20 million types yes
if you search hard enough you can find anything in the microsoft api 'docs'
maybe even a way to auth via pigeon
i would like the method with the most vendor lock in please
auth via telephone call by sending custom binary protocol in morse code beeps
does that suit your needs?
I did Google enough tbh
wow the cursor is now animated
I am reminded how much I hate in-app messaging
I wanted to be fancy and shit and add typing indicators and what not, now my messaging gateway takes ages to populate messages so it's not even like it's real-time anymore and typing indicator doesn't even work anyways
๐
are you using high tech level messaging system or basic one
Ima be real the frontend for messaging is a jumble of mess I hardly understand
I only know that now it actually works
as far as I know

The server gateway makes sense to me
but the client side shit goes over my head
more in regards to the use of tanstack with it since I use it for state
what counts as high tech level? 
have you used a high tech level?
I was thinking of using StreamChat
until I saw how expensive it was
๐
Rather struggle and build my own solution
๐
Okay so
I lied
it doesn't really work that well
Messages between users yeah, but I was setting up system messages and uh....yeah that failed
๐
Well that's another version of me problem it was a QOL thing not a necessity
i dont even know high tech one xD i was used to implement chat system just like how discord works. event based for everything real time, and fetch during initialization
how did you do it? did you use socket.io?
or some kind of websocket?
yeah socket.io, it was implemented on my social media app, which was when i was on high school
yeah i know ๐ i still love your social media app!
making it on high school was truly outstanding, i have no idea why you are not all over the news

they like to make everything viral nowadays
you deserve news coverage frfr
noooo ๐ญ not that pro. thereโs still asian kid under my age better than me
the asian kid under my age better than me is takiyo 
you are a few months younger than me so that counts
xD
which tc39 maintainer thought this was a good idea to add into js in the first place
they add this but they wont add decorators
mfs want to add memory management into js https://github.com/tc39/proposal-explicit-resource-management
now add rust lifetimes to js ๐
function something<'a>(s: &'a string) {
console.log(s)
}
yes yes
also add multithreaded programming with mutex, rwlock, reference counters, atomic counters, semaphores, and all that ๐
ditch the old event loop ๐คข
at that point you are just turning js into golang
Plz any one give me gift i lose all my

guys i kinda need help
im kind of new to development so i would like feedback from people
on my bot
is there any way to have test webhook of vote?
yeah theres a send test button
on the page where you set it up
it sends test one
it doesnt have a body
body is undefined
weird
I have no clue
bearing in mind I'm using old webhooks so that may be different
need to migrate tbh but I cba
on that note js actually has.. atomics??
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Atomics
this probably means were not too far off from better multithreading support in js
did you not know about this?
i have indeed not, i thought js was sticking to its usual single thread focus
SharedArrayBuffer has been around for ages
true
or use this https://github.com/sustrik/libdill
ignore literally any link being broken
this is essentialyl the same thing by the same dev with a working site https://libmill.org/
that link on their readme lmao
this looks huge tbh for c codebases
It can execute up to 20 million coroutines and 50 million context switches per second.
not sure how that compares to go tbh
i can imagine go will probably perform noticeably worse but probably not in a way that you'll notice (or something you cant improve within go) running a web server or something similar
depends how both are written really
bro istg i really hate xampp or the mysql, it keeps corrupting because i didn't stop it when i do power cycle
Lol
Maybe use your own setup using a docker compose file
here is a gemini generated compose file
# 1. PHP + Apache
php-apache:
image: php:8.2-apache
ports:
- "8080:80"
volumes:
- ./src:/var/www/html
command: >
sh -c "apt-get update && apt-get install -y libpq-dev &&
docker-php-ext-install pdo pdo_pgsql pgsql &&
apache2-foreground"
# 2. PostgreSQL (Database)
db:
image: postgres:15
environment:
POSTGRES_USER: dev_user
POSTGRES_PASSWORD: dev_password
POSTGRES_DB: my_app_db
volumes:
- pg_data:/var/lib/postgresql/data
# 3. Adminer (DB Management)
adminer:
image: adminer:latest
ports:
- "8081:8080"
# 4. Apache Tomcat (Java Servlet Container)
tomcat:
image: tomcat:10.1-jdk17
ports:
- "8888:8080"
volumes:
- ./webapps:/usr/local/tomcat/webapps
restart: always
# 5. SFTP Server (Open-source FileZilla Alternative)
# This allows you to upload files to your server via SFTP
sftp:
image: atmoz/sftp
ports:
- "2222:22"
volumes:
- ./src:/home/dev_user/upload
command: dev_user:dev_password:1001
volumes:
pg_data:
how does that happen??
mysql should absolutely not be corrupting itself after a power cycle ๐ญ mysql is acid compliant which means transactions either fully happen or dont happen at all, there should be no partial state after a power outage unless you messed some setting or OS disk write caching up
i dont know how well this is gonna go cause i didnt test it
i'm not sure, basically i just start mysql through xampp, do my development thing, and i shutdown my laptop when is stop using it. when i start the mysql back, the system database corrupted, i can't login to my created user, every action made has no result on the phpmyadmin, same goes when i try to connect via my app. then i backup the data, spin it up to create new one, and move my only databases and the ibdata, it works again. never happened something similar with postgres for year
nooooo ๐ญ
maybe you're using the older myisam engine in mysql? double check you're using innodb but that definitely should not be happening
the default is innodb yeah, altho the corrupted table is the internal mysql itself, which uses mix of innodb, aria (the most), and csv
where's the api option for my bot
you wonโt see it until your bot been approved on topgg
so I wait
yup
Hey i have a question. I have a aervice running on port 3000 on my vps. It is fully locked down. I need a service that exposes only port 3000 to a specific set of ip ranges.
I used ngrok but they are expensive to run now. Hence i was thinking about using cloudflare tunnels
What do you yhink?
How would my external server connect to it then?
With my public ip?
Is that safe? I mean i know which ip ranges connect to it.
So i can maybe set that using ufw?
yeah you can do that with ufw
At least from my understanding, using ngrok or cloudflare tunnels (unless cf tunnels has ddos protection) is just kind-of moving your issue further away
for example
sudo ufw allow from 203.0.113.0/24 to any port 443 proto tcp
I have done that before with my own network where I only wanted internal IPs to connect to ssh but allow public connection to other specific ports (chief server / management server -> managed servers)
unless public ip in your context is your own router for example
if thats a vps already connected to the internet, i dont see issues with ufw
i see, i will for sure use ufw.
for a security endpoint i used ngrok but they are ridicilous in pricing for a tunneling service. So i figured to just use ip port combinations and limit to a specific ip.
there's a problem though, the external server requires https. I could access ip:443 but the lcaol server runs on port 3000.
probably something portforwarding from 3000 to 443 could be possible
will check in a bit
hmm
or a reverse proxy or something
yeah that's what i thought of as well, proxying 433 to 3000..
i figured to use caddy
it provides free tls so it forwards any 443 requests to 3000
caddy works good for that yeah
and it's easy to setup lol one config file and that's it.
thanks for brainstorming with me Aurel
i sometimes yap here but that's my way of brainstorming
same, just not here too often
yeah butttt this is 144$ cheaper
ngrok is DAMN expensive lately. I remember that they were free.
I mean they still are but with ephemeral domains, no ip control. In prod this is not super safe.
the bright side
does anyone know a site for splitting up images into 2x2 so I can make a fancy big discord emoji out of 4 parts
nice ty
managed vps iirc means youre also partly paying to have their support team on standby to assist you with server tasks you may have, so that price would make sense albeit on the more expensive side
unmanaged is cheaper but youre on your own when it comes to any issues/things you want to do within the vps
yeah but then again this is industry standard, you usually only ever want to pay for the hardware rather than also for their support (but many offer it as an addon)
im not sure if this is still the case but contabo does occasionally help you free of charge for certain support stuff, but usually only if you claim theres some issue on their end
I use proxycheck to check if an ip is actively used in proxy or a vpn.
I was thinking of caching its responses for a day for each ip. But iโm not sure what to use to cache them so the retrieval is fast.
(Users that donโt get verified for the first time try to re-verify multiple times)
I currently use postgres for the db if that matters
that would immediately flag me tbf

How come
it didnt when i was working cause I didnt have my thing enabled but
Well it is ok Iโve noticed many people have their vpns on first time
But most of them turn it off
I mean
mines my own
so idk if it gets flagged
can send the IP for you to check if you want
If its hosted in a vps probably
dedicated server
U can check it in proxycheck by urself
i host my vpn on oracle cloud and it comes up as hosting, but not as a vpn
so you can still detect it though
proxycheck.io is a premium Proxy detection API with a full featured free plan. Try us today!
They have the check ip on their main page
Lol
yeah same as me
never a fan of vpn detection but if its part of your bots selling point then i guess go ahead
this is funky
I think i have set the risk to 50 or 60
So u might fo through
Im not sure if i should block hosting
please dont
Lol
no comment
well, self-hosted vpn is a legitimate use case
but so is self-hosted botting, and regular vpn
I was thinking that
plus if you are blocking regular vpn, then self-hosted vpn shouldnt be an exception
Right
hey guys, do you know what the average wait time for intent verification is right now? My bot has been stuck at 100 servers for 3 weeks.
It's Discord. It could be even a month 
i feel victim to typosquatting
luckily it was a fake package to prevent this
https://pypi.org/project/cryptography-fernet/0.1.0/#description
i got my first one in 6 days and second in 11 days
second was longer since it was during christmas
and what intents do you have enabled?
message content
painful intent to get
first one was denied
second was approved
hell yeah it happens again
2026-03-01 11:47:23 0 [ERROR] mysqld.exe: Index for table '.\mysql\db' is corrupt; try to repair it
***WARNING: .\mysql\db could not be opened: Error: 126
47% 58% 71% 84% 97% 100% (0.0 seconds); tables to flush: 1 0
(0.0 seconds);
๐ ๐
xampp says it all
tbh im clueless, probably some weird setting you have enabled somewhere or somethings weird with the environment youre running it in
havent used xampp personally either
well yeah fair my laptop dies due to overheat crash while i was gaming quite often, altho postgres survived all the crashes for months. iโll try another alternative or find the solution, thanks
why are you using mysql anyways? i thought xampp came with mariadb since mysql is a proprietary hellhole
(thanks oracle for ruining a good database)
oh, itโs called mysql on the xampp, although the driver/engine it uses is mariadb
sure i can use standalone mariadb but my uni use xampp on everything db related
tomcat and phpmyadmin is bringing me back at least 10 years
oh god
at my job we use c#, iis and sql server which i hate but given people are forced to use xampp in 2026 i think i have it good
tho sql server is a good database just too bad its proprietary
also im sorry but if youre paying for a university to teach you a 24 year old stack because they couldnt be bothered to update the course then youre being scammed ๐
might as well give you machines running on windows xp as well at that point
huh? they don't care which Java version we used for mine ๐
as long as it's Java 
muehehehe
:D thenk
at least make it less obvious youre trying to max lighthouse
wym ๐
lighthousmaxing
theres always gonna be someone who goes "this site sucks and is shit"
no lighthouse is gonna keep you from that
I mean iโm not trying to change peopleโs preferences
Tims statement on that reduces -50% of all lighthouse values
timshouse - takes the score from lighthouse and divides by 2
I'd end up with nothing :(
๐ ๐

guess they havent meteor-proofed the datacenters
๐
some people on reddit are saying it was a missile
lmao
i am bad at geography so i didnt think UAE would or should be involved in any of the conflicts that are happening right now
It very much is https://en.wikipedia.org/wiki/Al_Dhafra_Air_Base
Cyprus as well (RAF) so yeah, doesn't really matter how far or close
afaik even hotels got hit
brazillian news talked about how a group of brazilians got stuck there because their flight got canceled, and heard explosions while on their way back to their hotel

omg
today I learned! ๐
goated language feature
finally my bot got approved thanks jamie
thanks
Are you logged in btw?

i am and i have no custom instructions so its very confusing
llm quirk im guessing
deepseek does that to me often
This is scary
Its a matter of time before someone links up stuff and makes something big out of this
i dont think it actually pulled anything from other users conversations (at least i hope not), i think its just rambling that was included somewhere in its training set
i remember the time a while ago when chatgpt started spitting out a response to other users messages that wasnt mine lmao but that was an unrelated issue
Yeah
Its a matter of time before a huge scandal comes up with these chat services
Its so much private info
One fuck up and we are cooked
Companies have to start being more responsible with user data
some people feed these public llms their secrets and medical info and its scary
i believe that those llms mention that they might use the conversations you have with that in training somewhere
good point, for that then yeah buti meant more as in no live access to other conversations
that would make a lot of sense
you are basically using a single dataset which keeps on increasing when you interact right?
so i guess they are kinda using it live
and i even heard chatgpt starts its thought process even before you actually hit send. its starts from the moment you start writing
I'll outsmart AI and start pasting my text instead of typing it 
haha
you are still giving the data in the end
if you are gonna give it anyway you can just write it
there so you get faster response
thats both kinda smart and scary, but it seems feasible looking at the network requests, it sends everything you type to openai instantly without you hitting send
if you want you can remove all the event listners from the input field
looking more at their requests every time you refresh they also call this "prepare" endpoint which does some kind of proof of work???
what on earth are they doing behind the scenes
i have never seen an api do this
Interesting it was Chinese specifically, I know a lot of LLMs struggle more with Chinese Vs English due to the word structure
Wonder if they're feeding in more Chinese data to compensate
it only seemed to happen once, after retrying the response it didnt respond in chinese again, maybe it just happened to select a chinese token by accident
llms work off probabilities for next tokens so sometimes it can select a token that isnt necessarily the best prediction in an effort to diversify the output/prevent duplicate words
maybe what happened here
true
soon
If the product is free, you're the product
-Sun Tzu
yayy
Hey guys i want button in embed how can i not able to locate this in djs.
You can't do that in embeds. You can use container component tho
https://discord.builders/ is very useful for playing around with components v2
Use new Discord components in messages for free and stylize your server however your want.
i think we should dig into this
and publish our findings
chatgpt website isnt using browser inbuilt input feild or text area they have a div setup and is capturing your keyboard
We should call the cops tbh
Iโm sure this is mostly for improving their products. You agree to the same terms for basically every single product out there to โcollect data to improve our servicesโ
They are probably using it to train more models
google is definitely doing worse lmao
though on a side note i am surprised that despite google having access to so much data (google search engine, youtube, comments, reviews etc) their LLM models still arent as good as some of the competition like anthropic
I prefer Gemini pro over Claude for coding-related tasks
i just use the version that seems to be sufficient
I prefer Gemini pro for the first 3-5 queries, but Gemini seems to have really bad context windows despite saying that it supports a massive context window. It generally struggles to remember things Iโve given it from 3-5 queries ago, which makes it annoying to work with
Itโs pretty great for one-shotting frontend tasks though
I usually only use Gemini when I have documents I need to upload or if I am prototyping a frontend from a single prompt
i havent tried gemini much though admittedly other than the free model sometimes but since google seems to be running a google one free trial for one month i might give it a go more
I find that ChatGPT has the best long-context experience (at least from my personal experience with it)
im slowly migrating from chatGPT to gemini
Though gptโs web interface is horrendously laggy when you have long conversations, because despite paying every engineer over $1m TC, they still havenโt figured out how to virtualize DOM elements that are out of view
I have that issue on gemini with 1-mesage-chats
anytime it responds a large code block
4gb tab mem usage
GPT is even worse. I genuinely have to start new chats because after like 20-30 messages, the typing starts to lag and itโs impossible to get a timely response
Then I have to re-explain the chat/issues I was working on
Itโs insane that they still havenโt fixed it. Iโd imagine there has to be a reason for it, but I couldnโt possibly think of one
This is actually a pretty cool benchmark
Anthropic has been doing great work with their models in terms of hallucination %
Iโm surprised to see that gpt accepts so much bullshit
gpt is horrendous in terms of that
you have to double check every second message because it can just be bs a lot of the time
notable mention from yesterday
it corrected itself in the second message but why is it suggesting complete bullshit
claudes numebrs are impressive though
Yeah Iโve noticed that, especially when it switches from the higher models down to the free tier once I run out of the trial credits for the day
ive got chatgpt to make a nonsensical question and asked it to itself, gemini 3 pro and claude 3.6
chatgpt just ate it up, gemini pro gave a sort of joking answer but did recognise it makes no sense, claude straight it called it nonsense
gemini 3 pro seems a lot better with this but still fumbles sometimes
although claude did call me out here, maybe anthropic is explicitly training their models to recognise nonsensical questions??
because claude suspiciously performs miles better than the other llms
i wonder how it handles realistic questions that sort of make sense but dont
gpt be like: "ah the classic <insert random string> question/problem"
claude be like "you are absolutely right"
also, i a once again working with the most bullshit api ever invented, that uses GET requests like this:
/basicspacedata/query/class/gp/FIELD1/VALUE1/FIELD2/VALUE2/FIELD3/VALUE3/format/json
i would quit
i dont understand why some of these developers want to make their lives hard
thats what happens when developers are primarily scientists and not developers
Wtf what api
what's that even for?
should've guessed
luckily they have a yearly zipped data dump so i can avoid using their api for the most part
converting it all into an sqlite db
200+ million rows xd
sqlite goated
reminds me of the banana hamster meme
is there any way of setting up a premium based commands which unlock on voting every 12 hours currently i used the top gg webhook by ports etc ive seen some new setups like the vote tracker and different setups can they work somehow same?
cuz im switching hosting very soon which might have issue with ports and that method for me might not work anymore
Hosting providers usually have all ports allowed I/O
If not, you can look into the ufw command on unix based systems
you mean you are using your ports?
for incoming traffic
you should at least get 2-3 ports open
its kinda scuffed for me the hosting i use has ports but a 15% cpu usage and im thinking to change then the other doesnt have any ports says to buy premium
how much are you willing to pay for your vps and how much resources do you need?
i might know a quite cheap vps provider
so how much is it? ^
porlly like a 2usd or sum i dont mind paying is just they dont support payment method i use
i dont have a paypal currently so yeah
im so cooked
lmao
monthly?
yeah\
its from germany
works tell me in dm might be not allowed here i think promoting nd stuff
yeah
not all
okay I think I finally have a thing that makes me despise Java
it has no import aliasing ๐คข
its supersets have it but not the main language lol
no wonder people's class names tend to be hyper specific and long
wtf
why not
people's class names are rarely hyper specific and long, that's enterprise or legacy java codebases
i am concerned of my class names colliding with developers that use my library 
what's the lib abt?
it's the Top.gg Java SDK 
I have so many classes ๐ I was so concerned that maybe "Webhooks" is too broad so I decided to add "DIscordBotList" prior to it
ppl still use discord4j?
but well, I'd not worry much abt conflict, the chance of it happening is very slim
deadnaming topgg >.>
you can also just shorten it
so instead of DiscordBotListWebhooks just do DBLWebhooks
that's what I thought too, but the previous version still uses DiscordBotList
thought it'd be way too breaking ๐ what do u guys think?
the library's name still has DBL on it
i took a peek at libs.advaith.io and it looks pretty alive to me
if your doing a major version bump then it doesn't really matter if it's breaking
for the classes above it honestly might make sense to have those in separate jars and use capabilities to pull in the one you want
i can grab an example, give me a second
jarJar(implementation("dev.ithundxr:lotus-codecs:1.0.0-build.8") {
capabilities {
requireCapability("dev.ithundxr:lotus-codecs-minecraft-neoforge")
}
})
that basically imports a specific module from that library and pulls in the base library aswell
so sorta similar to cargo's features array
yes, that's what i am trying to do
I divided the library into six source sets
don't you think "Webhooks" is a bit too common of a name? ๐
yeah that's perfect
a bit yea
oh, ur adding support to rest libs too
mhm! ๐
oops, just realized feature names can't have dashes ๐ my bad!
i got a bit too used to Rust lol
๐ฆ
@quartz kindle what do I do with this
you could probably make a pretty nice jellyfin media server with that
@knotty night
thanks
weblate MIGHT be the worst open source software ive ever used
what the hell is that
im gonna install a working os on a real drive and try running ollama
I am praying my breaker will not kill me
Is that a blower fan next to it for cooling
it could be but no, not enabled




