development

[peak acorn]

To the point where i was thinking about buying a development server to run bare metal linux and just remote in to do all my stuff..

[fading osprey]

I have made my system as close to the dev environmment as I can, I also have a server that I do test most stuff on, the best thing for me was committing to using docker for as much as I can, because it makes that a lot easier

[peak acorn]

Yep I was kinda thinking of trying something like that

make development docker containers that have precisely only the packages that are required. Volume mount my actual local project and have the containers run off it

Only problem is I use windows and WSL is so brutally slow for file IO that some things really dont work well

rm -rf ./node_modules/ for instance... simply doesnt work from wsl

(afaik docker is hosted through wsl on windows)

But my buddies are trying to get me to buy an older workstation/server system and just remote into it to do all my development. Speed things up and create a more reliable experience if it works.

[fading osprey]

indeed it is, I use arch for my development stuff and windows for anything im doing that cant be done in arch for any reason (most of the case for me is gaming, a lot of my games dont play nice with wayland or linux in general, so windows is still a need for me sadly)

[peak acorn]

Last time I used bare metal linux, I just usb booted onto ubuntu... just to test something basic for someone, cant remember what it was exactly. Anyways, immediately, cursor wouldnt show up, any common fixes on google wouldnt fix it...

Reminded me that linux is to this day only useful if SSHed in from a server far away.

and no fuckin GUI

[fading osprey]

meanwhile im daily driving a notoriously quirky OS and havent had any issues since I installed it in December

[peak acorn]

Couldnt be me

theres no excuse for booting into any popular linux build and it immediately being unusable

[fading osprey]

well for a start you were on a live ISO, not an actual OS, so there's that, live ISOs arent exactly stable at the best of times (even with windows)

[peak acorn]

then why exist 😵

either way... stupid

i'd be happy to run bare metal linux on the dev server if i ever build it tho. Wont run a GUI if I can help it.

[fading osprey]

KDE is generally stable, Gnome is a joke to many, and the bible for others

[peak acorn]

Just no need for it if I had my way

Supposedly there should be some effective ways to use vscode on a host machine and run everything on a server which is what i'd want to do

common in big companies i guess

[spring cradle]

i love how thats the exact issue i ran into on windows last time

dont touch it anymore

too unstable for games

[peak acorn]

lol

well either they both suck, or we both did something wrong 🤔

[spring cradle]

everything sucks

the universe is hostile towards order

[peak acorn]

Real

[next igloo]

(MSYS2 MinGW64 CMake) My program is intended to be a stand alone exe. When the program is run, it will attempt to read from ./myProgram.config, and write to ./myProgram.config when the user exits the program.

Ideally the user could set a custom path for myProgram.config; So when the program starts, it will automatically load the config file from the custom path. However, I run into the chicken and egg problem, I can't store the custom path to the config file inside the same config file. Is there a way for me to store this data inside the exe itself?

[frozen flame]

On macOS you can use the defaults toolchain, i dont think windows has a replacement for that though

Maybe a file in a set location (beside the exe if it's not in the applications folder?) that has a pointer to the actual settings folder?

(A la symlink or just a path in said file)

[wooden shore]

You just need 2 separate configs

[frozen flame]

[peak acorn]

Two config files.
myProgram always tries to read file ./myProgram.configPath to find the config path.
If the file isn't there, create it, set it to ./myProgram.config, use that as the config path
If the file is there, use the contents of it to determine config path
Then read or create config file accordingly

[neon oriole]

as a nvim user i can objectively say , that it sucks bad and hard, is barely configurable , and not configurable by novices , eg you need a second editor to configure it in any workable way... but its the best we got and thats sad , since it defenetly has/had potential to become good (if they stopped thinking of themselfs and everything about vim to be the best possible and any other opinion is a person who has skill issue's or just isnt made to use vim, and actually look at the facts about what they got || and yes ive looked at the code for some parts and i would not want to fixs some of the issues either judging by what i saw but thats not an excuse to claim that it is by desing and that the desing is any good||

[peak acorn]

just use vscode 👺 👺 👺

[neon oriole]

haha lol , a fanboy,... well i just wrote this to make it somehow less painfull to maintain a consistent keymapping that is not completely idiotic, now my mappings are stored as ini files...

[peak acorn]

yea i have zero clue what im lookin at :)

im just a lil babby i dont use vim 🤡

[neon oriole]

mapping to maintain: ```ini
[cursor_move_1ine-up]
CMD : <Up>
<C-i> : nvxi,noremap,silent
i : nvx,noremap,silent

how you do keymappings in nvim land : ```lua
vim.keymap.set({'n','v','x','i'}, '<C-i>', '<Up>', {noremap = true,silent = true,desc = 'cursor_move_1ine-up'}) --cursor_move_1ine-up
vim.keymap.set({'n','v','x'}, 'i' , '<Up>', {noremap = true,silent = true,desc = 'cursor_move_1ine-up'}) --cursor_move_1ine-up

anyone defending the nvim way over any other way seems like wanting complexity where there is none, but the first is identical to the second code block in functionality , only nvim people seem to think the second one is the best possible way of doing things

not saying my way is the best but atleast its better

[spring cradle]

seems like wanting complexity where there is none

idk if the defaults are that complex

sure its not the most intuitive, but its not like it takes a whole homebrew external program to maintain

idk why you keep paining yourself by using nvim lol

[frozen flame]

the kind of people who use vim are the kinds who's opinions can usually be ignored tbh

[peak acorn]

I havent looked into it, but I cant imagine I would actually edit faster in vim than vscode

Supposedly vim can give you a lot snappier language tools tho..

[solemn solar]

Counterpoint: What you are used to and what brings you the best features is what you like best
If someone never webdevs, they don't need insane CSS and HTML integration, but for someone like me that might be necessary if I wanna work fast, but I have no need for docker because I don't work with it

[fading osprey]

So as someone who has used with little exception almost every major editor, ide, and cli for writing code.

They're all about equal in terms of most things, each does something better than another

• IntelliJ Family - best in class database management tooling built in, High quality code suggestions and recently optimized
• Visual Studio is good at, idk, something I guess
• Vim is good for snappy responsive tooling in the terminal
• Nano is good when you need nothing but highlighting and ability to do something quick
• VSC, is better than Visual Studio at everything it does
• Atom - lol, it's dead move on already
• Eclipse - What are you actually doing with this?
• Emacs - Just use Vim, stop torturing the carcass of your editor.
• Brackets - Eclipse but worse
• NP++ - Windows Eclipse but worserer
• NP - get a real editor
  Outside of that, there are not many left that are major editors.

Really everyone falls into three classes, lightweight and performant (vim, nano, np++, etc), Heavy and bloated but good QoL features (IntelliJ, Visual Studio, etc), and somewhere in-between (vsc, atom, Zed, etc)

[solemn solar]

I see one code editor missing:
pen and paper

[fading osprey]

Leave, and do not come back, you are no longer welcome here

[shadow goblet]

imagining the whole code in your head

[solemn solar]

That's the best because you aren't constrained by any hardware or software, you can make your own hardware and software with your own features

[frozen flame]

VSC can absolutely fall into bloated lol

Rust analyzer moment

[fading osprey]

Not out of the box it can't, which is what I'm getting at

[frozen flame]

It's useless for anything other than webdev out of the box so there's that

[fading osprey]

My point stands though

[grizzled steeple]

VS in my experience is good for one thing and one thing only:
Windows C/C++/C# and Windows .NET development

[fading osprey]

My own limited experience of c# would have been much more enjoyable in Rider tbh

[solemn solar]

In my limited experience of only using VSCode, I haven't had any complaints so far apart from no proper CSS class linking in HTML and perhaps an update breaking Vibrancy Continued, but it's really nice
Also it's free

[frozen flame]

Rider is a much nicer experience imo

Once you disable their weird custom annotations thing anyways

[neon oriole]

(while i never used editors like brackets extensively , if i have to list the ones ive used and include the ones i discarted after a minute i think my list would be 3 times longer) because of the natur of my daily drivven distro i'm quite dependant on a terminal editor , and as such i used to use JOE, but that one has been dead for a while. (missed that one in the list btw) , pico ,nano, and micro just lack the ecosystem of (n)vim, and dont have the functionallity build in like joe(wich because of it neads less of a n eco system) , emacs really is its own meme: great os but lackluster editor . i also think you are wrong abtout Eclipse (maybe not wrong about perse about eclipse itself but), i recommend LiClipse (for beginners with python) Visual Studio (especially the older versions) is good at : Creating WYSIWYG Visual Basic Tools. as a gui editor i use Sublime 3 and 4 and kate allot as editors unfortunatly they lack (i thinik ) the possibility to follow variable and function calls acrosss files for many langs. For C i mainly use KDevelop, for PyQt Gui's (i always build gui's with pyqt even for not python programs i make the gui with python and qt) i use ERIC7, I used to use Aptana3 allto for PHP/html/css development. i Stilluse Notepad++ on linux because of some of the plutgins that make life much easier for doeing certain things. i do allot of python stuff with pycharm not because i like it or think its great (most of the the stuff one would use an ide for i actually do by hand outside of it because i dont like the way pycharm does it) but my brain is verry fast and good at filtering out colors i dont want to see, and i put in the effort of writing my own (based on molokai-neon i think) colorscheme , that combined with a plutin that consistently colors variables and rainbow brackets , i get verry colorfull code that i can find stuff in verry fast and i cant seem to get that in another editor as easely

and to defend my nvim tool thing: the gui tool I just adapted from something I had, the main thing I wrote for nvim is a conversion tool from ini to lua and back that also keeps track of overwrites, so if i map the "i" key to be the <up> key it warns me that I overwrote the goto insert mode functionality of I , if i would also overwrite <insert> to mean something else , i get an error that insert mode is no longer bound..., and indeed the lua way is not the only way of mapping stuff, you can also use the :noremap way but the lua way is the only more or less consistent way that works for anything for mapping.
And yes you really do need a second editor to make the mapping for vim, its that or Also memorize the default keybindings of vim to edit the configs in vim, which obviously is an oxymoron if you're trying to remap the keys you now have to learn first. if there was a logic to them or if they were any good (the default mappings i mean) i could overlook this but as it stands, while mappings are subjective the default ones are objectively bad... and are only the default ones because of legacy brought over from times there were no ISO/standard keyboards yet

[fading osprey]

what do you mean, why dont you want to <ESC><SPC>fe<SHIFT>h to see your .env files?

[neon oriole]

i also give allot of free passes to vim already for stuff it could do but i get why they dont or i myself would find a chore to implement in a way it works everywhere vim currently works. like being able to map any key on a keyboard , and even detect key-down key-hold and key release , or map two identical wor two combi-keys, like pressling left and right shift simultaniously or left-shift -> right shift to got insert mode or right->left shift to go to normal , wich is possible technically just not feaseble to do on every platorm and env where vim runs

[neon oriole]

you assume that that means anything in my editor , bad news .... <ESC><SPC>fe<SHIFT>h has no effect in my nvim

[fading osprey]

thats also the other half of the problem

no matter what you are doing, chances are someone has it bound differently in their config

[neon oriole]

exactly

and you can find functions that they are bound to internally in theire code for most things, but not for everything

[fading osprey]

yup, not to mention this is before you get into things that are just not built in by default. God forbid you play with package managers like lazy.nvim as an example, because then your config will just 10x in complexity over where it already was

[neon oriole]

duplicate line can be done with commands and not using keybinds that coud be mapped otherwise using: <Cmd>co<CR>["x]p<CR> but thats not the case for everthing

because of the cmd, wich cant be remapped and , <CR> wich isnt the enterkey really its a virtual key that cant be remapped i think if im not mistaken

or <Cmd><A-s>confirm quit<CR> to quit and ask to save if modified

[fading osprey]

<CR> as a concept is also handled differently by every OS, and even to some degree, different distros and flavours of said OSs, Windows use <CR><LF> for line terminators, but in most *nix OSs it simply a <LF>, which makes the <CR> hard to trigger unless you specifically map something to it, but as you mentioned, that is often difficult to impossible

[neon oriole]

haha really , but thats only halff the job , god forbid you use a plugin that has no snippet to paste for lazy .... your translating code for one pakagemanager to another packagemanager and hoping it works and gets loaded in the right location.... or it might break stuff and next time you start vim, you might end up starting up default no config vim

[neon oriole]

ah yes but thats ffor every editor, i meant using <CR> in the command as a key (wich technically does not exist) works in every possible nvim config - i think - , as a linux user i can see why unix uses \n but i do think the windows manner makes more sence

[fading osprey]

I ran into this issue the other day, the one I was having to do it with was presence.nvim which proved fruitless since it doesnt work for me anyway lmao

[neon oriole]

stuff i tried to make it doe that should be really simple, some of it i proved to be really simple just to make a point, is terrible in nvim or even impossible , ive tried many different distributions over the year but they all are impossible to change ... so then i tried nvim kickstart, but before starting it , i threw out what i did not want or need and added my own keybinds ... bad idea , i also dont liked its here is an installer file that needs to remain part of the config just for making stuff work approach , i found kickstart-modular, liked it better since its split over files but god try removing stuff from a really barebones thing even is painfull. but on the otherhand its the only editor that is somewhat updated regularly, so if you want support for a language like mojo or zig your choices are limited to editors that get updates, ...

[vestal crest]

so. work have just paid for github enterprise... and have said we all need to create github accounts so we can be invited to the organisation... told my manager that i already have one, can i use that... "yes... but please... please.... remember to set your real name and work email address as the global name and email address on your work machine... i REALLY dont want you to get kicked from the account and i dont want anybody complaining to me that they think we have been hacked, every time you commit something" 😄

[neon oriole]

(was looking for the thing i wrote just to disprove the vimfanboys , but forgot the name, just remembered it, its the most commonly used excuse by vimfanboys: that is because of terminal limitations not becaue of vim: so name is termlimitsmyass :D)

[fading osprey]

im indifferent on vim tbh, I use it, I dont hate it, it has its quirks

[neon oriole]

haha i do use my real name an mail for github.. but id have a different problem, my personal code is litterd with the following variablenames (used as foio and bar but i mostly use : test , and... ikkel, wichj might look bad ) but https://www.youtube.com/watch?v=8tTCGz2E4JI 😄

[peak acorn]

ctrl+p .env enter

🙂

[fading osprey]

on my editor that does nothing, but accidentally fat fingering ctrl+W+S at the same time will split my window vertically

[peak acorn]

vscode

not sure how to split window on vscode 😔 just right click lol

[silk eagle]

https://tenor.com/view/yoda-there-is-another-star-wars-jedi-gif-5140031

https://en.wikipedia.org/wiki/Microsoft_FrontPage

all you need to know about it is in the first sentence, HTML editor and website administration tool

oh and and and dreamweaver

[limpid knoll]

hey no leaking what i use to edit my webpages

[frozen flame]

NO

[neon oriole]

over the years:

• vi(m): vim nvim helix kakoune oni oni2
• pico: pico nano micro
• Joe : Joe
• Crimson
• Atom
• Scintilla
• pycharm
• Thonny
• Geany
• gedit
• Nedit
• jEdit
• UltraEdit
• kate
• kedit
• kdevelop
• ERIC
• Liclipse
• Qbasic
• komodo
• Notepad++
• Scite
• Sublime
• Aptana
• Qt Creator
• Qt Designer
• VisualStudio
• Dreamweaver
• FrontPage
• Maquetta
• netbeans
• Altera
• Quartus
• Visual Basic (5,before it was part of studio)
• mono develop
• DevC++
• codemirror (but ill list this as the only web editor but used this one allot for webdev as part of modx/pydio/ajaxplorer)
  ... prolly forgetting quite a bit from the time i was still in school..

[neon oriole]

i got that one in shool :p and frontpage , i used aswell... the revelation when you realize you can section a webpage using tables instead of frames ... i mean if you used frontpage you made layouts using tables , must be

[midnight wind]

modern dreamweaver is just webflow

[neon oriole]

when i got tought (well tought dreamweaver not html since i knew that already) dreamweaver , it was still macromedia dreamweaver :p

[fading osprey]

Yup, same here, but I've got Dreamweaver to thank for me learning how to do some basic html stuff

[sudden cosmos]

I was at peace with html briefly when I was using Brackets to do some little stuff, then into the fire with angular 1.5 in webstorm 😬

[peak acorn]

in highschool they had us do dreamweaver

terrible,l ol.

[limpid reef]

Macromedia (take that, Adobe!) Dreamweaver and Flash 8 were what I started on too... until I very quickly learned how much extraneous garbage Dreamweaver added to your code in an attempt to "make it more compatible / compliant" - then I switched to EditPad Pro, followed by Notepad++ for now.

To be fair, Dreamweaver's "edit your code, insta-refresh to see the results in real time" was kinda slick at the time, but these days completely unnecessary when the browser or dev environment can do that on their own.

[reef star]

remember guys, always close procmon after you are done 😄 Pretty sure my memory was not 55% utilized, but close to 100

[twilit elk]

Had similar issues recently on later Windows 11 insider builds where the virtual memory was slowly yet surely eaten up by some memory leak or something somewhere..

the memory usage would be normal range but virtual memory would slowly be eaten up - despite what size I put ( I mean just slower, but in like 24h-48 it would go from like 16% at boot to 97%+ and then stuff would start behaving fun)

[haughty oyster]

I've had the service responsible for the programmable F12 button on my HP laptop bug out a few times and slowly but steadily eating more and more RAM. Once I let it run and killed it after around 10 minutes at which point Windows had assigned 75GB to the page file with the 32GB of system memory also being full. PC building sim has also had some issues where going to the store and going to any category would make it freeze and eat up all my RAM until the page file got to around 100GB and then it would crash

[neon oriole]

haha , i liked the templating stuff , then i discovered <?php inlude ?> and that was better than anything dreamweaver could do :p

[neon oriole]

on windows i always replaced the default tasmanager with processhacker2 was more powerfull and much lighter than the default taskmanager, and it could start programs as system (kernelspace >> adminspace)

[reef star]

I am doing my development most of the time over ssh, so I felt no need, until now 😄

[peak acorn]

Anyone use a home server for development?

My buddy was trying to convince me to buy some older server hardware, get like 40+ cores and 64-128+gb ram. Supposedly itd be faster than a 13900k almost definitely (I have a 12600k right now). That would let me do all my development remote from my windows desktop and get a full linux environment that has really nice multicore performance for builds and parallel tests and whatnot.
its kinda tempting, but im curious if anyone has experience setting up their development environment for remote development and if it was a pain to set up or if its finnicky at all

[fading osprey]

ive got an old HP 1U server in my garage which is used for some development stuff, got a github actions runner on there and such like, its handy to have but by no means necessary. as for the second part of remote dev, I do all my dev within Arch anyway, so its irrelevant for me but I have heard it isnt half bad with vscode

[peak acorn]

I just really cant do windows on my actual main machine, just cant. So moving on from there and finding workarounds for issues.. I have been using WSL for most tasks that cause any pain in windows. And honestly that usually solves my issues immediately.
Git is all in wsl because I think authentication makes so much more sense. Half of the project is built on my linux side.
But theres a problem. File IO is crazy slow in WSL. Like, git status takes up to 10 seconds on a project that isnt even that big.
And something that operates on a large number of files, like installing packages, can also be prohibitively long.

Thats the main reason running a server for development would be beneficial. But then it also gives me the added benefit of being able to optimize for multi core performance on the server, and my main PC can remain relatively focused on gaming, which would be pretty sweet.

[fading osprey]

I just dualboot

[reef star]

That depends, what kind of dev work you need? Ihave an ESXI host (rip) up and a few virtual machines, for testing etc

[peak acorn]

Typical full stack dev environment. So theres database, backend, front end, and some extra services running. Those are either in docker contianers for some testing, or just run locally which is more common.
My whole system (running chats, browsers, etc) use like 40gb total.

But yeah the major slowdown I have day-to-day is running test suites and compiling code and building images, and git operations, even though its only a few seconds wasted per, its annoying lol

[frozen flame]

couple seconds
https://xkcd.com/1319/

xkcd: Automation

'Automating' comes from the roots 'auto-' meaning 'self-', and 'mating', meaning 'screwing'.

[peak acorn]

nnoooe
not "I should automate"
rather "I should buy a $5,000 server to marginally improve it" 😈

[frozen flame]

Well yes but there isn't an xkcd for that particular case

[peak acorn]

haha ik ik 🥸

[neon oriole]

the only time i ever used WSL2 , i did not really notice wheter it was actually much slower.... (and since i ran gentoo under wsl, it had to do allot of compiling so i think i would have noticed.... https://wiki.gentoo.org/wiki/Gentoo_in_WSL

[peak acorn]

in my project linux git status is 2.8 seconds
windows takes 0.34

time git status and Measure-Command {git status}

[neon oriole]

$ cat /etc/wsl.conf
[interop]
enabled = false
appendWindowsPath = false
``` ?

[neon oriole]

was repl to this, seems to work for some people

[indigo dragon]

Is this on an NTFS file system or a Linux file system?

The Windows<->Linux filesystem overhead is substantial

Also I had a bunch of permissions issues at some point which is understandable

[peak acorn]

NTFS, files living on windows machine.

[indigo dragon]

Yea try it on a Linux file system

[peak acorn]

im pretty sure it is faster, but then it makes running commands from the windows side nearly impossible so its not rly an option

File paths and stuff are super jank

[indigo dragon]

Yea there is no good solution beyond having one clone point to the other as a remote

[peak acorn]

yeppy

or i just build a large fast server and run it all on linux and remove into it for all development 🤯

[indigo dragon]

How big do you need?

[peak acorn]

I mean just 64gb ram and tb or whatever of ssd
But if I do that I want to get a beefy cpu setup. Something thats faster than my desktop would be by a decent margin.

[indigo dragon]

Very cool

Yea if/when my side project stuff scales to the point where other people need to use it ill probably throw down some money for a server and have people VNC/RDP into it

[wooden shore]

Meh I think the remote development route isn’t worth it. I’ve got a server and I’ve tried it. I just use wsl

[peak acorn]

A pain to setup or continuous problems after the fact?

[wooden shore]

Just continuous problems, can’t remember exactly what it was.
And just wasn’t worth the hassle, much easier to just do it all locally.

[peak acorn]

good to know

[wooden shore]

What are you building exactly? I have a couple year old machine and I have no problems building large .net, c and rust projects

[peak acorn]

project isnt actually that big
React front end
Go backend
Postgres database
And a few "micro" services

[wooden shore]

Are you looking for a server to host those services on or a build server? Kinda sounded like a build server at first.

[peak acorn]

To generally develop locally on. Including builds and running tests.

[frozen flame]

Ah fuck I meant to reply to the message above that one

[grizzled steeple]

I'd say it's more that drive that is born to corrupt and not the filesystem...

3 out of 3 Seagate Ironwolfs I've bought died within a year of usage lol

[molten oracle]

True lol. I've experienced the same with ironwolfs, but barracudas have gone on fine for decades

[reef star]

let me guess, you've got a 4th now?

[grizzled steeple]

I was specifically talking about the ones I bought. First one I got a replacement which I still have but no longer use. Of the other two only one was replaced while I got my money back for the other. That replacement drive I sold without ever even plugging it in...

I've since replaced those Seagates with Toshiba drives

[peak acorn]

Something that has tormented my go developing for a long time now.
Say a function generates a number.
the function internally uses a function that can error.
So you gota do something with the error, most likely you just want to let it bubble up by adding an error to your return types and passing it in.

So you do keep the int return as an int and provide a default value, or do you do *int and make it nil?

Kinda feel like nil *int is the way to go. Because if you just do an int with default -1 value or smth they might try to use that value without checking the error.
But I feel like you should trust that the caller will use the results properly.

So then, maybe the return type should assume the function works correctly and you do int. And let the error be the signal that it was.. an error.

[indigo dragon]

I do all writes through pointers and only ever return a status

[peak acorn]

in go (afaik) its pretty common to return pointers of data you create in-function. The GC lets you do that

Just an easy way to make return values nullable

[neon oriole]

anybody a clue how to deal with these not occurring , or get removed automaticly? (eg all new not old)) git inserts these all over my project: (about 50 files total spread over folders , so i dont have a onesize fits all aproach to removing them unfortunatly)```git
<<<<<<< HEAD
<<<<<<< HEAD
def run():
wgt['Fnx']'Run'
wgt['App']['Fnx']'Run'
return

    def run(wgt):
        wgt=gnr.Clean(wgt)
        wgt['App']['Fnx']['Run']()
        return wgt
    return run

def Fnx(wgt):
    wgt['Fnx']={}
    wgt['Fnx']['Configure']    =        wgt[w['name']]['Fnx']['Configure']
    wgt['Fnx']['Add']                =        wgt[w['name']]['Fnx']['Add']
    wgt['Fnx']['Main']    =                wgt[w['name']]['Fnx']['Run']
    return wgt
def Init(wgt):

<<<<<<< HEAD
<<<<<<< HEAD

as you can see theyre not in the right spot either since i suppose this would be the correct positions: ```git
<<<<<<< HEAD
<<<<<<< HEAD
def run():
wgt['Fnx']'Run'
wgt['App']['Fnx']'Run'
return

    def run(wgt):
        wgt=gnr.Clean(wgt)
        wgt['App']['Fnx']['Run']()
        return wgt
    return run

<<<<<<< HEAD
<<<<<<< HEAD

def Fnx(wgt):
    wgt['Fnx']={}
    wgt['Fnx']['Configure']    =        wgt[w['name']]['Fnx']['Configure']
    wgt['Fnx']['Add']                =        wgt[w['name']]['Fnx']['Add']
    wgt['Fnx']['Main']    =                wgt[w['name']]['Fnx']['Run']
    return wgt
def Init(wgt):

[indigo dragon]

You can use -X to prefer one side in any case of a merge conflict

-X ours or -X theirs

It would be cool if somebody made an AI-based merger conflict resolver

[neon oriole]

yeah im a little scared that that will result in the wrong thing in some cases, since the placement of the head thigns is wrong and includes code not part of wat was differetn, in this case it would work out , but is that the case everywhere ?

[indigo dragon]

If it’s anything more complicated than all-of-one-side I don’t think there’s an automated way to do it

technically you could split the merges file-by-file, rebase at the end to squash it into one commit, and merge that without conflicts back to the main version (hopefully)

but that's still all-of-one-side, but you've just split it across multiple merges so you have file granularity instead of no granularity

[neon oriole]

yeah its all one sided but as you can see on the location of the <<<<<<<<<<HEAD things in the first its below a part that is not changed between the two , in this case if i would go with all original , it would delete ```py
def Fnx(wgt):
wgt['Fnx']={}
wgt['Fnx']['Configure'] = wgt[w['name']]['Fnx']['Configure']
wgt['Fnx']['Add'] = wgt[w['name']]['Fnx']['Add']
wgt['Fnx']['Main'] = wgt[w['name']]['Fnx']['Run']
return wgt
def Init(wgt):

offtopic dont know what torvalds was thinking when he made git but,... changing user files and not storing an exact copy of the original just possibly a diff is not great practice 😦 , (because it assumes bitflips arent a thing,neither is filecorruption)
not that tat would solve the problem but it would allow me to restore without problems and pic a different route to combine the two branches

[indigo dragon]

i think as a principle not assuming data integrity is unreasonable for any project

unless it directly works with the hardware (like a file system or maybe the memory subsystem)

[neon oriole]

not assuming or assuming is unreasonable? , my logic usually goes like this , if i deal with files that aren originally created by my program or could have been modified outside it by the user.. you dont mess with it , you duplicate , and change the duplicate archiving/renaming the original next to the new file (some.conf -> some.conf some.conf.backup.001

[indigo dragon]

assuming bitflips exist in most programs is unreasonable

the hardware should work well enough to not have these issues

if it doesn't and you believe this is a problem, you either need to shell out for ECC or some other solution to the problem at a lower level

[neon oriole]

true i only mentioned it as a more general term of corruption ,since i didnt want to say assuming your code is perfect and wont cause the corruption,.. but its linus torvalds were talking about so :p

[indigo dragon]

yea i wish more platforms had ECC and i was talking with somebody who insisted it cost "literally pennies" to enable this on any platform

[neon oriole]

intel dont wants it ...

[indigo dragon]

i didn't tell him he was wrong but i asked him some probing questions to get him to realize that

[indigo dragon]

ehh it is more complicated than that

OEMs also don't want to shell out any more money than necessary

[neon oriole]

and yes the cost of ecc would be the same as the normal ram , and enabling it wont even cost anything if the normal stuff is no longer needed to be made

[indigo dragon]

and most people i know primarily use laptops where i've never seen ECC used ever

[indigo dragon]

you need to store the parity data somewhere

and the whole idea of a QVL becomes relevant here

[neon oriole]

the extra cost is more in having two pipelines one ecc and one not , than in the extra cost of ecc in itself

[indigo dragon]

I mean this does add another RAM chip and RAM chips do cost more than literal pennies to insert

[neon oriole]

and the extra cost of ecc would end up being less since instead of small group for ecc now , big group for normal now, to one really big group of ecc then. wich reduces productioncosts significantly

but also margins that could be made in markup

[indigo dragon]

that's true but what percentage of the customer base is interested in consumer ECC?

[neon oriole]

and companies need ECC so they are willing to pay for it no matter what -> huge margins, while if everybody used ecc they could not ask those margins and be competitive

[indigo dragon]

i think it's a pretty reasonable assumption that most people who know about ECC and care about it would just go for a Xeon outside of some subset of enthusiasts

i think most places would just save a buck and not shell out for ECC

[neon oriole]

nobody cares but would you argue the same for seatbelts? nobody knew they wanted it until they got the benefit of it

[indigo dragon]

or the places the ECC is required are already served by Xeon

[indigo dragon]

any safety critical system should rely on ECC existing

[neon oriole]

if you realize what is vulerable to bitfilps in a computer , havving ecc sould be a matter of lets have one spot atleast center to it all that has correction logic, every place where firmware ,cache , is located, now the cache for a DAC having bitflips might not be as critical as the cache for the fpu , or the cache of a drive before writing the contents to magnetic eg but... it insane , and the hotter the thing runs the easier the filps happen , ask the google datacenter in belgium ;P

[indigo dragon]

oh yea i'm aware

i'm just saying consumer chips are built for the average person in mind and the average person IME doesn't care

[neon oriole]

same goes for seatbelts in cars , or lets say tcp , since most consumers would accept udp only if that was all that was offered for free or cheap ,and using tcp would require spending double on a computer ... , and we know they do because they accepted bluetooth and that has no security nor has it any package order or integrity chechecking , they even accepted wireless paying , where everyone near can sniff the transaction with no effort so , they will accept whatever is offered, but if ecc was offered for everybody as the only type , nobody would be complaining about that, and allot of errors that exist today would not

[indigo dragon]

seatbelts in cars and ECC in consumer chips aren't really comparable IMO

there's a very clear upside that's obvious to the public that seatbelts reduce fatalities

and I believe at the time the NHTSA already existed as a regulatory body here

(at least in America)

the Nader Raiders did a good thing

[neon oriole]

i would have gone with your standpoint a few years back btw untill i saw this one and (another one about how ecc works and having studied electronics realizing the cost of it is neglectable) https://www.youtube.com/watch?v=yQqWzHKDnTI

[indigo dragon]

actually nvm I think the NHTSA came after

[indigo dragon]

my perspective isn't that it shouldn't exist, my perspective is honestly it should exist but it won't take off because of incentives

i can see a future where Intel just does memory on-package and that reduces the complexity/burden on OEMs for ECC

and since much more of the process is internal, the OEMs don't front as much cost and it's easier to adopt

[indigo dragon]

regardless it's pretty clear that safety regulations are a responsibility of the government in general

this actually begs a question of what's the cheapest computer with ECC in any form factor

x86, ARM or RISC-V

dusts off his military-hardened space-certified 8086

[peak acorn]

Might as well just overwrite everything with yours lol

gonna end up confusing ppl by deleting their shit either way

[indigo dragon]

yea i'm not saying it's a good idea

but it would be neat technically

[peak acorn]

it could b cool

but you always gotta human review and merges are like THE time you need a human to review every last thing

[indigo dragon]

yea

[peak acorn]

🥳

im working on some AI document classifier and scanning and we require human reviews for everything, this just helps them do it a bit faster

[neon oriole]

i still think ECC for all ,... if everything was ecc the only extra cost would be payed in time, by the consumer (since ECC is by definition a little slower) ,what i actually wanted to say and i cant believe im saying is : fixed the git problem but thanks to pycharm... and its undohistory that also tracks all files in a similar way to git without it being git, what i mean is everytime you run the tests or an external change happened and you opened the project again, it creates a snapshot that can be used to rollback the wole project ... wich is nice in this case , back to the old rebase instead of merge and it seems to have worked

[neon oriole]

that is untill a human finds a way to automate controlling the AI code changes ... 😦 and auto reprompts the ai for corrections at infinitum till the tests succeed and controls no longer error out... then they ask the ai to customize and write those controls for every project ... 😦 that being said i dont think we should classify llm's under ai, its artificial yes , but its not inelligence , its just lossy textcompression where the compressed version of thousands of files is stored and controlled by offsite corporations.... intelligence is never perfect so i dont considder something intelligent untill it can go past mimiking , and or ask clarification because it has contradicting or multiple possible directions to go in, so when asking i need a function for my program that returns fibonacci numbers , should not show me code since it does not know what language my program is in. but ask for it . but since its just lossy compression of text it cant even contemplate a whole text to evaluate it contents as a whole in order to determen if its credible, if using stack overflow as dataset, all the code examples that dont work are equally part of its possible nets , but without the information attached to it that that wont work or was an example of something wrong...

[peak acorn]

Well first I dont think it matters if you call it AI or not. AI is a moving goalpost so its really complicated.
But you need not worry about the rest... Its definitely possible to get language models to ask questions first and then solve problems.

It's even beneficial (this is what my project is doing right now) to get AI's to type out their "thoughts" or do other processes before giving you a final answer and it actually improves the results. its pretty nuts...

What do you make of that?
I can ask the AI basically this:

You are given an unstructured dataset describing a living being, your job is the classify it into the kingdom that it's species resides in.
first, list clues from the data that might help classify the creature
second, develop a diagnostic reasoning process that will help come to a conclusion about the creature's kingdom
finally, use the clues and reasoning process to come to a final determination
your answer MUST be in JSON format following this example
{
  "CLUES": ["Fur", "Four legs", "Social"]
  "REASONING": ["Animals are typically the only kingdom that has social behavior, legs, and fur"]
  "KINGDOM": "Animal"

And it simply performs better than a straight question for most tasks. It also helps corralling the AI into proving formatted data

[neon oriole]

i have this one somewhere but i dont know if its still working and can't test it atm since no system with agp that still works anymore :p '

[neon oriole]

haha yes but that would have to be done programaticly, and thus is not ai, in order for it to be ai it has to generate the more info question out of its own not because of programatical nudges...

[peak acorn]

Certainly that will come with time

[neon oriole]

as i said , llm's are really just lossy textcompression , that reconstructs the original text according to its best guesss or approximation

yeah but the process for how it works would have to change dramaticly the way it currently operates cant do that ever. but this way does allow for maximal controll by its owner so

[peak acorn]

I am curious what you think your brain does when you're coming up with words and code 🤔
Sure theres more going on, but at the end of the day do you not think it's a complicated web of compressed data?

Our brain has little simulation processors, language processors, crazy complicated memory schemes. The brain even kind of separates these processes

[neon oriole]

and i do think it matters that if marketing is calling something what it isnt , as this might be harmless mostly or easely controllerd but by naming it this way when the real deal comes around , everybody will be oblivious to it... and think its still the same thing where it most certainly isnt

[peak acorn]

Yeah thats probably fair

Im very very curious ( and scared of course ) whether "true" AI will come around gradually or suddenly jump from code completion/video generation/vision models to boom superintelligence

[neon oriole]

mm t his would suggest i saw all code i created before hand. and i never figured something out myself, like changing little stuff untill it worked , py def function(**k,b='s') nope py def function(*k,b='s') nope py def function(b='s',*a) nope py def function(b='s') def function(**k) this works but ... ```py
def function(*a,b='s')

[peak acorn]

tbf LLMs can almost certainly generate code its never seen before. By merging together different things its been trained for.

So obviously you can too

[neon oriole]

so my complicated web is not dependant on input in order to creat e output , as it can generate input internally, and use that to generate partial outputs internally verrify or fallsify , manipulate them before creating output, most importantly my complicated neural net in my head is not based around tokens

but around concepts

[peak acorn]

You don't think LLMs can do that?

I dont think the product is actually good from what i've heard, but have you seen DEVIN?

It has a command line and browser that it will run and test code and (sometimes at least...) solves problems based on errors that it's own code gives it

[neon oriole]

if they are based on neural nets , in the same scalable manner the popular ones like chatgpt , no they cant , they always require input before they start doing anyt hing, and they stop doing anything once output is generated

they dont contemplate what they just created , and ponder if there might be a better way... animals like crows (whom i do considder intelligent) and honeybadgers,dolphins,eg do

[peak acorn]

So just code an input loop where it simply
PROMPT => LOOP { generate answer => send to "done?" model => feed back } => output

What does it mean to comtemplate though?
If it means to use your current best answer to direct future answers, LLMs can do that...

Do they do it perfectly? of course not. Are they limited to a context windows of only a few megabytes/kb? yup.

[neon oriole]

now your programming again and that s cheating that is cheating, but you would have to continiously loop, but even then , the loops should be part of the neural net , not from out to in , but from one layer to the previous layers and so forth but that makes them les scalable

[peak acorn]

How is it cheating. Its supposed to be artificial, basically by definition its running a computer program

an LLM isnt even a single pass through a neural net y'know? It goes through a program tokenizer, it constructs data structures to store 'attention' data or however that works... and only then starts running a series of neural networks

[neon oriole]

its cheating because you have it generate output , then by hand feed it back , always, not by choice of the program, as in , i am sure of this no need , or mmm this answer looks suspicious i'll better recheck...

[peak acorn]

i mean..

If chatGPT behind the scenes fed its input back to itself multiple rounds over and didnt tell you, what then?

if programming behaviors is cheating then the concept of an LLM is inherently cheating

[neon oriole]

this answer makes logically no sense, its full of fallacies...my source data on the subject is wrong, so i should not use it as source material for this subject anymore

[peak acorn]

ChatGPT actually runs moderation models before the data even reached the actual GPT model, is that cheating?

[neon oriole]

wich it cant objectively discard sources

[peak acorn]

Explain that again plox

[neon oriole]

no thats just like the fbi listening in on the phoneline and cutting it when we start talking about something were not allowed to ...

[peak acorn]

lol mmmhm

What about RAG?
Connecting a language model to a database where it can request relevant data and be retrieved for a second-running with the included data in context.
How is that different than your thoughts?

[neon oriole]

if you were to feed it a fairytail, as sourcematerial, and you ask it is it safe to use wood to construct your home... and it answers no cause a wolf can come and blow it over and eat you, it cant considder that wolfs cant blow that hard and thus in future construction questions discard that fairytale as accurate material to darw from. in similar fashion if mostly trained on religious text, it cannot list all logical fallacies in those text , because it does not know what its own sourcematerial was, so if you feed it communist propaganda , it doesnt realize that that is propaganda and only usefull as examples of communist propaganda not for its actual content

[peak acorn]

That just sounds like youre complaining about what it's trained on and thats pretty much irrelavent to intelligence

[neon oriole]

or try : wait 5 hours beffore answering this question with '5hours have past' dont say anything until you think 5 hours passed ...

[peak acorn]

If you literally lived in fairy tale land... Well you too would think that a wooden house could be blown over...
Or not even that

If you are literally 5 years old. Its very possible you actually think a wolf could blow your house over. Are 5 years old wholly unintelligent?
of course not, they just arent as smart as adults

Also like, if believing religious texts knocks you out of the running to being intelligent 😳 rip to like 5 billion ppl

So i mean that cant be used as an argument i think

[neon oriole]

you can replace the subjects with any relevant topic ,... if you have it ingest alien horror romans , it will use the test in those romans but it cannot assess that it prolly is not the best source for answering questions about spacetravel

[peak acorn]

so?

[neon oriole]

eg it has no ability to assess the material it was trained with, nor does it know what piece of text comes form wich source and how credible is that source for subject x

just saying , lossy textcompression is not intelligence ,... now at some point in time artificial intelligence will exist for sure , but this is defenetly not it. it lacks every aspect of what we normally attribute to actuall intelligence

like it can imitate a c compiler pretty well but if asked questions for c code it does not choose to run it past its internal c compiler to see if it even compiles ... out of its own , just to be sure not givng code that does not work

and that makes openai the worst name for the company imho :p its not open and its not ai ... 😛

[peak acorn]

I mean yeah thats a totally separate question

openAI is literally a profit seeking organization now which is ripe for abuse

[neon oriole]

haha i put it there because i just realized that :p

[peak acorn]

But I just think your definition if AI is quite wild. Why is it a problem that it has modules put together to form something that solved problems? Thats what we do

We could directly hook up a C compiler and bypass the whole lossy conversion part

Why is that not part of it's system? it totally is

[neon oriole]

dont know how watson was programmed or googles alphago but i think they were closer in the right direction for ai then llm's are

[peak acorn]

1. The issue of training data making a model "not based in reality" is not a factor for intelligence. If you actually in fairy tale land that would be your reality and you would still have intelligence
2. Adding modules to a larger system to improve its capabilities is not "cheating" to create intelligence, as artifical intelligence is inherently a mixture of systems

[neon oriole]

but llm's have a finanial advantage because of more data , and more cpu power means betterr llm, so its really scalable

[neon oriole]

assessing it would be ,

[peak acorn]

AFAIK alphago is basically an advanced search algorithm that uses many metrics and neural networks to decide how to search the problem space and come up with a move

[neon oriole]

nobody ever had to tell me that i cant use comic book situations for reading situations in real life , i kind of found that out myself. nor did anyone ever have to tell me polititians arent updating there knowledge of anything because they learned it ,but because it gets them more votes so parroting them is probably not a good idea to find out the truth

[peak acorn]

I have no idea what that means

[neon oriole]

assesing sources and where they apply and where not

without the need of someone externally assessing them for me.. even if many would like have had to

[peak acorn]

But you do need external information to assess it. You used your eyes and other senses to over time determine that comic book superheroes arent real.

So if you were not born in our universe, but rather in a white box with nothing but books, how could you possibly know? Would you not be intelligent with just books in your void because you dont know whats real to us or not?

BTW: What I am trying to get at is yes LLM's aren't really "AI". They probably wont be the sole source, or many even any source, of "AGI". They are not sentient and an LLM can never be.

However, there is not really any reason to think that an advanced LLM system can't "reason" (if we can even define that)

it's also very hard to define intelligence in the first place, and therefore hard to define artificial intelligence. However, what I'm trying to point out is that all of your reasons to discount their "intelligence" are faulty

[neon oriole]

if i were to ask you : is brown a color? and why not? or anyone with basic understanding of language and a little intelligence will not give this as anser to that question , most will either contest the question, or reconsider their knowledge about colors and if there is a spot in there for where brown is not a color... and if found give that. they can also see that the first question has an inferred answer and thus needs no answer because of the second

[neon oriole]

it cant reason... because it cant update its knowledge

[peak acorn]

It can update its knowledge, it can poll the internet and grab up to date information!

[neon oriole]

reasoning requires that new information previously unknown can be added to the pool to draw knowledge from and reassessing ones presumptions if they still stroke with the newly drawn knowledge

[peak acorn]

An LLM system with modules like internet search capabilities can absolutely do that within its context

[neon oriole]

and most importantly update the presumptions if t hey no longer are correct with what your knewly found knowlegde tells you... and no having it access the internet does not make it the same since it cant update itself with that information , and they cant allow that to happen because it cant score the validity or value , correctness of what of the internet you have it read, , havving it read from the internet is nothing more than adding that to the prompt that goes to the input, without you havving to look for it on the web yourself

btw this is not what some intelligent thing woulld answer on the brown question btw

[peak acorn]

That's not what an intelligence creature would answer?

Didn't you say non-human animals have some level of intelligence earlier?

[neon oriole]

i atleast would have expected , no brown is not a color , because every color that exist is in the rainbow and brown is not in the rainbow or in a prism

im sure it has that information somewhere as part of its dataset

but if all you see is tokens and the probable token to follow it up with....

[peak acorn]

I think this is simply an entirely unreasonable opinion you have.

1. An LLM will easily override its built in information if prompted with updates
2. An LLM CAN AND DOES update its own information with the internet. There are existing tools that allow the LLM to make calls to the internet and feed it back into its context.
3. An LLM does not generally need to think about the validity of a internet query. You might disagree with it, but google is generally pretty good. Hell, you could just give the LLM known good sources, would it become intelligent then?
4. Once again, why is the system feeding data back into its own prompt something that discounts its reasoning capabilities? This is literally what humans do all the time. We have a short term memory that gets updated with things we think about and things we read.

[neon oriole]

1. no it does not , if not provided as its input... since the entire prompt is put as its input each time it can just assess given that input , it internally does not update, it also does not include that input with a new prompt so it does not remember , as such there is no point in "teaching" it things using the same manner as promptting it ... it has to be taught differently if teach is even accurate for what actually is happening.
2. same deal no the llm does not update, what it got from the internet just gets added to the input prompt , if it was updating using the internet , it would remember across sessions.

[peak acorn]

How is updating the prompt different than "updating"
it is literally the deciding factor on what it generates.

Just because the weights and biases arent being updated, does not mean the system is not changing

[neon oriole]

3. yes it does unless you explicitly ask it for the content of the internet reinterpretted , if the top results on google are propaganda and propaganda only it needs to assess that and look for a piece that actually has the credentials and no fallacies in it to reply the correct result and not the propaganda

[peak acorn]

Once again. If falling for propaganda means you are unintelligent, then there is exactly zero intelligence humans alive...

[neon oriole]

okay exam math, here is a paper with all the formullas to use , or here is an exam you need to use what you know and only that , now as a human after the exam you prolly remember some of the formullas the llm wont unless you provide it the paper again...

[peak acorn]

Why is the LLM not allowed to keep the old exam in context?

[neon oriole]

the values and triggers of the neural net do not change to reflect the new information

[peak acorn]

Why is it not allowed to use tools like MemGPT to literally store it in a database

We are not just talking about an LLM in itself. We are talking about an entire "intelligence" system built upon an LLM and numerous other modules. Just as humans work.

[neon oriole]

as human you have no choice the neurons in your brain will change with everything , the neurons of an llm are static so.... and you can by thinking change the structure and connections of those neurons , and by figuring out answers to questions, no matter how manny questions an llm solves it wont ever change itself

[peak acorn]

The neurons of an LLM are not the only factor at play

Plain and simple.

Like lmao, sorry but what the hell are you even talking about

[neon oriole]

what else is there ? i mean neither for a human we have eyes and senses but those arent required for our brain to do anything ...

[peak acorn]

gpt 4 LITERALLY HAS WOLFRAM ALPHA BUILT IN
no, you don't get to just say "the ai isnt allowed to use calculators" it is an "artificial intelligence" system. It will get literally every math problem you give it correct better than you.
Sorry but you make absolutely no sense and your arguments for why LLM systems will never be intelligent dont make sense.

I dont have a great definition of intelligence but you keep coming up with random arguments as to why its not, and all of those fall flat on their face

[peak acorn]

There are databases attached, there is context.

[haughty otter]

https://tenor.com/view/popcorn-guy-relaxing-popcorm-popcorn-gif-26404903

[neon oriole]

maybe different approach , i have an llm on my pc , there is no way to turn it on. but to prompt it and it will turn of the second it generated input,... now i can repromt it and it will turn on again , but unless it is provided with the previous prompt it wont even remember that it happened , let allong what was said , so it either gets an answer or it gives you a wrong one no matter the previous conversation,... if i did the effort to make it see my point on a topic , and it aggreed with me in the end, i have to provide it with that whole prompt each time , where if the structure changed it would remember , maybe not the conversation , but the result of it and it agreeing with the new knowledge ...

[peak acorn]

Reasoning does not require 24/7 processing.
Even if it did require 24/7 processing. You could force an llm to constantly run inference and boom you solved your problem.

I am done with this convo 😔 sorry love ya bud but holy cowe

[neon oriole]

nope but i already gave an answer to the what reasoning requires. and why it cant (yet) since it cannot update the state of itcurrent knowledge it cant reevaluate everything it knows about the subject and for a new opinion as the core of its next statement , simply because it does tokens , no concepts so there is nothing to reason with on the side of the llm , ther is just the next token given this input and the lossy compressed texts that is my learning data, is mostlikely to be ... , the fact that its output can be a reasonable answer does not mean any reasoning happened internally
btw i just gave chat gpt a problem to solve, here is a text but its cypher text , a simple conversion 1:1 to english letters and then a simple transposition should give a meaningfull text in english... it took a while but it gave me : (as meaningfull english text:
R-ŋRAᛗW.GIHEIYI-MæYWX-æN-/YEP-GææD.IZDIᛋJ-ŋLRJ-ÞY-/DMᚫŋ-ŋGI-ᛇJAI.EJIPJ-ÿI-D/MᛗŋHIKOI.XEMK-ÞY/A-JAEY.IZ/SIAEIMDI-ÞY/A-CFÿ.K-MAEJ-IKEJ-A-DLRWI-ÿEP-GææD.ÆA-YINIPPROI-DÆMJREMII-ŋŋEEM.IEÆZIA-Y-ŋ/ACP-æA-YIEA-MᛡNᛄIAIᛈ.XÆA-RHH-E/P-PRDAIK.
okay that was maybe a bit much but it could have noticed that that does not mean anyting. so i gave it something simpler 1 simple english word as ᛞᛁᚢᛁᚾᛁᛏᚣ it gave me as meaningfull word DIUINIITY not only does that not mean anything it is wrong as the place where it gives II as corresponding to ᚾᛁ wich are different symbols however since it knew it shoudl be an english word , it could have seen that it was wrong and just did its thing a little longer, and got it or it could tackle the problem in an intelligent manner , and use the dictionary to only use words to match against of that lenght , where characters 2,4,6 are identical and mostlikely a,e,i,u or o leaving not even much of a code to decipher to get to Divinity this does not require much skill not is it hard , but it does require some reasoning and intelligence

now i dont expect intelligence to be correct , i do expect it to know that its not right

and stop after getting R-ŋRAᛗW as no way this is english and thus my method is wrong

a crow trying to get a nut out of a jar with a stick will only try 2 or 3 times before realizing that this wont work with the stick he has , and either gets another stick or fashions a stick that would work for the task at hand ... , if a stick that would work is out of reach it will even try to use the first stick to get to the other stick ... but thats next level intelligence, llm's dont even realize the stick theyre using is the wrong stick and will not work

[acoustic pine]

i cringe so hard looking at my first website's CSS

600 lines of pure spaghetti

[solemn solar]

You would probably not like looking at my first code, or my current code

I had AI do half of this and ALL the CSS is in the html file

Current code isn't THAT bad but it still could be better as I often overspecify stuff still
(Though I have once referred to width: 100%; on one of my properties that shouldn't need it as a drastic measure)

[acoustic pine]

I'm developing a new website with Flask and MySQL
pure HTML and CSS without any UI framework
and actually going to have it properly designed now code wise

my old one is bandages upon bandages

trial and error without knowing the fundamentals on how things work

Tried dipping with bootstrap but i find it more confusing than vanilla css

[solemn solar]

Happens
I rewrote mine like 5 times before I even got to the point of finally settling upon one code setup, I just kept rewriting it because it sucked in the past

Also tbh nothing wrong with pure HTML and CSS, I use it and it's pretty nice tbh. Just gotta not specify everything everywhere, though my brother calls me crazy for doing a full website with just that (and Javascript)

[acoustic pine]

I'll probably tryout React when i have the time

[solemn solar]

Good luck and let me know if it's any good when you do try it, I've been thinking of using something to simplify the process

[indigo dragon]

This doesn’t look so bad

[neon oriole]

hehe might wanna watch some of these beforehand, (full disclosure, i hate everyting js and js related, and i stopped using it after jquery , so ... that being said of all the js stuff out there react is prolly the one i hear the most resentment of having picked it for projects)
so I would say: dont do js , increase your knowledge of html5 (barely anyone seems to know it and or use it as intended) and defenetly CSS 99% of stuff people do in js can be done with CSS in a much simpler and easier and defenetly easier to debug manner.
so if you break out that js lsp editor maybe think about can't i just do that in css: if its animation or interactiveness :https://codepen.io/YusukeNakaya/pen/ZadZxL defenetly look at this one sometime :https://codepen.io/juliangarnier/pen/nxPoNK not the empty js field inhere.
here is one i did at some point : https://jsfiddle.net/hvf10bL5/ to prove animations owere easier in css ...

the primeagen on the otherhand would say use : htmx , i cant argue against that ofc ,... he works at netflix i dont so :p

[acoustic pine]

ah you have a separate software for the 3D CSS
i almost going to blow up if it's actually hand written one by one

[neon oriole]

what i usually do is generate the css with php so , also that way i dont have to include all possible css, but i can reduce the size significantly to whatever the browser reports to be, (no need for a vert layout on a desktop, but does require some in small window consideration css so)

not for large images but for icons eg , ofc i use svg if possible , but i inline the xml of the svg, so the whole page including grapics is done and returned with a single html request,... usually is much faster and takes load of the server dealing with connections... , used to do the same for css ages ago but then i figured out that certain things are better cached locally 🙂 where css is usually limited to a single file (php generated so) icons tend to be allot more , and have more benefit from inlineing them, if not svg , i use base64 for the png's or jpegs or gifs....

and its as fast as using javascript if not faster due to no libraries and nothing running clientside , only difference is that if systemsounds are enable on windows you hear a click sound where whith javascript you dont :p

[red mulch]

Lol, "AI".

LLMs are the dumbest clever person ever.

The current design of "AI" systems are absolutely super cool and very capable. They are so close to "intelligence" that you can see how far away they are.

[knotty pier]

hello can somebody help me with Netlify and CSP Dynamic Implementation? Thanks

[acoustic pine]

AI is very very good at solving basic problems since there's a lot of data on it

It all breaks down if you give it any thing more niche or complicated

It can't even finish a basic logarithm problem i gave it

And any code longer than 20 lines breaks down entirely

[acoustic pine]

Yeaaa seems like circumstances got me pursuing web development instead of android

Reason being money

Currently using flask but I'll probably need to move to a more popular back end solution to be more competitive

Fuck that means touching JS

I've been avoiding it for a while but it's inevitable

[indigo dragon]

for some reason i'm getting a deadlock trying to lock stdin/stdout/stderr in rust

no idea why this is happening

[fading osprey]

If you've got any code that's locking elsewhere, make sure it's letting the lock expire before you lock elsewhere, I ran into this last week with a multithreaded app that never derefed a mutex

[indigo dragon]

yea i get that but i'm only ever locking in a child process after a fork that execs shortly after

so there shouldn't be any cases where the parent process has the lock acquired

[fading osprey]

Hmm, that's a new one

[indigo dragon]

may have found the problem but it would be weird if this is what broke it

the close(stdin_.as_raw_fd()).unwrap(); used to be close(stdin().as_raw_fd()).unwrap(); and it may have been trying to re-acquire the lock on the call to stdin()

but the code stops dead before that ever runs and the lock should be acquired later

i do lean pretty heavily on the drop semantics in Rust for functional correctness so I should probably know these things better

mostly around message passing and dropping receiver means the sender should stop

anyways that seemed to fix it which is weird but i'm not complaining

[indigo dragon]

also PDEATHSIG is GOATed and more people should use it

[wooden shore]

So many unwraps 💔

[indigo dragon]

yea at some point i'm going to throw a fuzzer at this and fix all the unwrap issues

or prop testing or something like that

my biggest issue right now is the tests take a long time to finish since i haven't bothered mocking out things that are slow

[indigo dragon]

also in many of these cases there's nothing i can do to fix the issue and nothing I can really do to propagate it up

since this is all in a forked child

the best I can do is abort with an exit code and catch this in the parent (which i already do)

[solemn solar]

Is there a good way to automate a process for scanning 300 images with Google lens (or similar) to find if the image is taken from the internet or not?

[agile wyvern]

not really development persay, but I want to make a custom version of the js based web game slowroads (https://slowroads.io/). I would like to download it and run with vs, and just change some textures. Is it possible to download the code from the game? Thanks. (obv it will be kept private and not reproduced)

[peak acorn]

slow roads is closed source

[next igloo]

I have an application published to Itch.io and GitHub. I put a link to the GitHub’s issue page for any bug reports, but I haven’t received anything there.
I think it might be that GitHub is too complicated for the average user, although 40 downloads isn’t “popular” enough to conclude that for certain.

I thought of using Google Forms as an easier way for people to fill out bug reports. Do you have any other suggestions/ideas of what I should use for users to submit bug reports?

[peak acorn]

Doesn't itch.io have a comments section?

[hybrid arch]

hello, I'm tryna figure out how this person did indoor mapping for this AR indoor navigation app: https://github.com/Gebort/FESTU.Navigator
there's a nodes.db file which looks like this: https://srcb.in/RcNaVVZUbN
how do i make a similar db for my building

[peak acorn]

For long functions

what do you people think about putting a table of contents at the top. lol.
Yes I know a long function typically means that you should try to break it up. But thats what im doing right now and its kinda nice to work with. Cos u can always skip to the top of file and then use the comment to figure out where to look inside the file

[untold badge]

Does 4800 MHz vs 5600 MHz RAM matter for any dev tasks on a laptop?

[keen sorrel]

It would also depend on the timings for most workloads

[untold badge]

So spending extra for better 'quality' RAM is worth?

[keen sorrel]

Only to a certain point. The performance difference isn't all that much even in the best case, but it also costs very little extra. I'm not familiar with DDR5 SODIMM at all to know where the ideal point is though, sorry

[untold badge]

Okie, thx for your input! It's a $40+ difference so just want to make sure

[acoustic pine]

are there a better way to do this

[peak acorn]

gonna get sql injected lol

[acoustic pine]

I mean it's just a movie list thing

Nothing to gain when one of the feature is to literally show everything

[frozen flame]

Not really the point

All SQL databases have a method of sanitizing data, instead of formatting strings. Never ever use format strings in an sql statement

[keen sorrel]

It looks so wrong, but what they're using there is the sanitized formatter from the mySQLdb python library. I'm not familiar with that library so maybe it's sanitizer is problematic or something idk. Though if that DB is 100% dedicated to movie list data, then it doesn't even matter since a cursor is read-only.

[robust dagger]

^ plus it’s pretty common nowadays that new programs are built on frameworks that has built in sanitization

[dim cedar]

Where should i start if i want to learn programming?

[solemn solar]

What kind of coding

[gilded stream]

Start with C

Or java

[gilded stream]

Where will you even host this

[dim cedar]

idk anything just want to get into it

[gilded stream]

C is a good language

Go for it

[molten oracle]

C is bad

[gilded stream]

Nope he can learn memory management and basics with that

[molten oracle]

rust ftw

[gilded stream]

Oh come on really rust

[molten oracle]

Safer way to manage memory

[molten oracle]

Weak typing, Memory leaks, null pointers are real problems that nobody should have to keep in mind while learning how to code. But that's my personal opinion.

[fading osprey]

I am known for my rust shilling, and I personally say it really depends on what @dim cedar is wanting to do. What are their goals with development

you wouldnt tell a web dev to learn C, that isnt how it works, they get nothing from that experience

[molten oracle]

Objectively, most people learn with python nowadays anyways.

[fading osprey]

you also wouldnt tell an embedded systems dev to learn JS, for the same reason

[gilded stream]

Well yes

[molten oracle]

Technically js works everywhere.

[fading osprey]

it really doesnt, I say this as someone who is fairly proficient in JS

[dim cedar]

im not sure about the sphere of it that i want to do what language would be the most overall good

[molten oracle]

It's not the best choice in most places. But it most definitely has tools for most purposes.

[molten oracle]

There is no "overall" good language, because the goal often limits the language(s) you can use.

[gilded stream]

I think that if a person gets used to it first then it would be really really easy to learn other things or languages or frameworks that are upcoming

[fading osprey]

for general learning the basics and figuring out what you want, Python is a good starting point, or Javascript, which is able to be used both on the server side, and the web side with minimal change in how the language works. Once you've figured out what you enjoy more doing, you can choose to focus your skills in on something much more advantagous than JS or Python, like C/Rust on the server side, or JS, for web, or Kotlin for Android and Swift for iOS development.

[fading osprey]

learning C isnt a requirement to be good at programming, I know many great devs who have the ability to pick up new languages like there's no tomorrow who didnt bother learning C or its (horrific) memory management principles. Language brigading isn't going to help newcomers to the space at all, we just need to provide logical entry points for the type of development someone is interested in.

[molten oracle]

Most of the popular languages have a garbage collector (other than C and rust, ironically enough) so memory management is not as high importance imo.

It's definitely a very useful skill, but not one that is universally required or relevant.

[dim cedar]

ill probably learn python then

[fading osprey]

If you need any help, pop a message in, someone should be able to assist 🙂

[gilded stream]

Should a upcoming dev go for c# then because it got everything (multiple frameworks)

[gilded stream]

Alright then start learning python here to help

[molten oracle]

A solid choice, it's also got a very positive community so I'm hoping your journey is fun!

[molten oracle]

C# is still used in 2024?

[gilded stream]

Some people do use it XD

[fading osprey]

This is like saying should someone buy a Suzuki Jimney because its got everything (multiple wheels), its a dumb argument, you use the right tool for the job. You dont shove the square peg in the round hole. C# is a very square peg for what is pretty easy to see as a round hole.

[molten oracle]

I just see go, rust or ts everywhere. Or C/C++ when it comes to systems/games but I don't really look much elsewhere.

[fading osprey]

Being a good dev is not about brigading a language, it is about identifying the right tool for the job, and learning to use the tool effectively for the tasks at hand

[molten oracle]

Tho TBF, doing stuff in languages not intended for the purpose is a fun side activity.

And might I daresay, VERY fun

[gilded stream]

Just like heking

[fading osprey]

indeed it is, but, not for a beginner

[molten oracle]

Well, yes. For a beginner, all I recommend is a want to build something.

Then deciding upon appropriate tools among the ocean of options.

[fading osprey]

exactly

Python is a pretty good all rounder language, it just sucks at any one specific thing

[molten oracle]

I mostly see people struggle because of not having the "want to build something".

When you're working towards a goal, you understand the purpose of each line of code way better than "just trying stuff out"

[odd crow]

hey i am in grade 12! i am learning python and i already know most in my syllabus so i am bored any suggestions?

i am really into development

[shadow goblet]

do projects

[peak acorn]

learn software design patterns

[odd crow]

What kind of?

[odd crow]

Oh that sounds fascinating

[shadow goblet]

any/all

[peak acorn]

https://refactoring.guru/design-patterns/catalog

Some of these are basic, some are more complicated, but most of them will be super useful to know.

Understanding how to implement factory/builder/singletons, and iterator/strategy pattern are probably the most important

[fading osprey]

additionally, you could try some leetcode challenges, theyre always fun ways to figure out what skills you actually have, and what skills you think you have

[next igloo]

I made a Google Form for users to fill out bug reports in my program. Do you think this is readable/understandable by the average user, or should I change/add/remove anything to make it simpler?

[solemn solar]

As a partially average user, this makes sense to me, might be a bit hard for a grandma-level user, but otherwise pretty good

Learning Javascript and I've made a simple transition thing, is there a way to make the code better? Seems very complex for a simple task, it's my first actual attempt at javascript by hand so it kinda makes sense, but would like some input on how it can be better

Essentially there are three buttons and I need it to be made so that once one button is pushed, the line next to it extends, and when another is pushed, the same line retracts and a new one extends

[gilded stream]

Btw I want to develop an os where do I start

[gilded stream]

Put bug report on top not on the bottom

[fading osprey]

There's a great tutorial series for Rust by Phil Opperman, (I think that's his name) called "Building an OS in Rust" or similar, it's not particularly high level os building, but it teaches you the stuff you need to know for certain

[marble rivet]

I don't know if it's considered developing but, I want to get into 3d printing, any recommendations for a beginner friendly 3d printer and 3d modeling and slicer softwares?

[acoustic pine]

Use a JS framework/library for animations

Vanilla JS will be convoluted

[molten oracle]

I would recommend checking out osdev wiki, it's a great set of resources.

https://wiki.osdev.org/Main_Page

[molten oracle]

There's essentially 2 options here.

1. Use an animation framework/library (I prefer anime.js)
2. Divide it into CSS and JS. Instead of applying and removing each style manually in JS, you can group them together in CSS and then it's significantly shorter to just toggle classes in js.

[molten oracle]

(here's how I would do it with Vanilla CSS/JS)

[solemn solar]

I'll have a look, thanks!

[solemn solar]

And for now I aim to be pure vanilla JS as I wanna learn it fully before going to a framework, not only taking advantage of the framework but being able to swap back to vanilla should the need arise

[molten oracle]

That's fair, frameworks the code shorter but modern vanilla js is pretty decent by itself too.

[acoustic pine]

I'd move into typescript

[molten oracle]

The only thing I would recommend is switching to TS

[molten oracle]

Literally what I was typing LMAO

[acoustic pine]

I definitely need someone else with actual artistic talent to do the UI design

Then i can just implement it

Because damn my designs is definitely "it just needs to work" kind of engineer approach

[molten oracle]

I just try random things and go for "does it look good" approach. I'm not much of a designer either lol.

I just end up with somewhat decent designs cuz I'm friends with some designers who start roasting me in DMs for every design fault I have.

[solemn solar]

Would knowing Javascript before starting TS be better? I still wanna just learn the core 3 since I already know core HTML and CSS

[molten oracle]

They are essentially the same but TypeScript has static typing so you get way better intellisense in most text editors.

[solemn solar]

I see, so knowing JS would mean I also know TS?

[grizzled steeple]

What's nice about TS is that if you don't want to specify a type, you don't have to. It'll try and guess the type based on initial value assignment. Not a very good practice to just rely on that, but it's easier to get started with it that way

[acoustic pine]

That's a lot of non type strict languages tho like python

[acoustic pine]

Easy enough you can immediately code in TS without prior knowledge of it

It's the usual basic C inspired syntaxes

Oh IT IS basically JS

JS with types
Typescript kek

[vast reef]

I use hostinger with an mail account, and twi alias.
I connected it with IMAP to Outlook, but how can I use my alias in Outlook?

[molten oracle]

I would say the other way around. Typescript is a little more strict.

[edgy socket]

I am not sure if this is the right place to ask, however, i am trying to build a small Webhosting environment that requires Apache, PHP, PHPMyAdmin, and MySQL. I and building this on Unraid using Docker Containers.
I currently have the following 2 Dockers installed:
ApachePHP8-Docker
MySQL

I thought that ApachePHP8-Docker would also come with PHPMyAdmin; however, gloriously it does not...
I am not very Linux savvy, therefore, I would like to inquire the following:
Is there any way to install PHPMyAdmin INTO the ApachePHP8-Docker Container?
And if not, how do you install SSL Certs to the "phpmyadmin" docker?
I ask because I attempted to use that Docker as a work-around however it refused to let me log in citing that:
"Failed to set session cookie. Maybe you are using HTTP instead of HTTPS to access phpMyAdmin."
And it has a banner across the top of the login screen saying:
"There is a mismatch between HTTPS indicated on the server and client. This can lead to a non working phpMyAdmin or a security risk. Please fix your server configuration to indicate HTTPS properly."

Is there anyone that can assist in this?

[peak acorn]

cursed font

[peak acorn]

on this

[solemn solar]

Definitely cursed

[peak acorn]

Is there any way to install PHPMyAdmin INTO the ApachePHP8-Docker Container?
Yes, create a custom Dockerfile that is based on your Apache image, install phpmyadmin there.

And if not, how do you install SSL Certs to the "phpmyadmin" docker?
Either in dockerfile copy, or via a volume mapping, expose ssl cert files (you could self sign if they just talk to each other)

[molten oracle]

Just because it's not one of

• JetBrains Mono
• Ubuntu Mono
• Fira Code
• Consolas (also cursed but win default)
  Doesn't make it cursed.

As long as it's a mono spaced font that is not comic code, it's fine.

[peak acorn]

No that font is comic as can be lmfao

[molten oracle]

Cursive with ligatures ≠ comic

[edgy socket]

I am relatively new to linux, do you know any good video or step by step instructions on how to do the first option? its easily enough to install Docket images from the Community Apps on Unraid, however, I have never built my own customer Dockerfile....

As for the second option, Im not really sure one what you are meaning, My PHPMyAdmin acts like it does not have SSL files, I would like to use Lets Encrypt and just let it plug them into PHPMyAdmin (if that is at all possible) to save on hassle.

Sorry if i sound like an idiot, i am used to working with prebuilt LAMP installations for this. However, I am trying to learn more Linux and Docker images "seemed" to be a nice and convenient way to deploy a quick Webserver for testing needs... Any assistance will be appreciated.

[peak acorn]

This might be useful?
https://youtu.be/SnSH8Ht3MIc?si=s9hGMew5Ru0DOrcl
I only clicked thru it briefly, but it should kinda get you an idea of how to set up a custom docker image.

For the ssl stuff. I'm not actually sure how phpmyadmin works. But presumably you just download some sort of certificate files and point your config to them?

[edgy socket]

Thank you, I appreciate it. I have been trying to find a place that i can directly download new SSL Certs to plug into my PHPMyAdmin system but I can seem to find a free site to do it.. They all seem to use automated systems that you interface with via scripts on CLI.. Ill try to continue to investigate that. Thank you

[peak acorn]

Yeah not sure how letsencrypt and stuff work on that front.
If your containers are only talking to each other privately, you could self sign and generate keys yourself. But if anything is publically facing you'll want to use a service

[edgy socket]

Yeah, ill want to use a service, this webserver i am making is a test bed for an HTTPS Request Hub for a system I am building in Second Life

[lofty igloo]

I just got a samsung tab s8+ and im very impressed with its display. Is it theoretically possible to write a program/driver for the tab to act as an external monitor for my computer and display 1440p 120hz and hdr with very less latency? my laptop and tab both have usbc 3.2 port which seems to have a speed limit of 20Gbps so can it be used to transfer the video like an hdmi? Many apps accomplish this kind of like samsung second screen, spacedesk but they are wireless and have refresh rate and resolution limits with unusable latency. The closest thing I have seen is "super display" which has unnoticable latency when not many things are moving in the screen but starts to lag when im moving a full window for example

[acoustic pine]

Definitely not
Mobile SOCs here don't have the capability to process direct video input (only some niche SOCs like those in PIco Neo 3 VR headset can do) and you are always required to use Encoding and Decoding

Also, HDMI 2.1 is 48Gbps

and DP 1.4 is 32.4Gbps

You will always have some latency

spacedesk supports USB btw

[frozen gate]

when your CI pipeline YAML is > than the src its building 😢

[molten flicker]

Docker is so pointless ngl

NixOS already solved all those problems two decades ago

Just Nix, actually. Not even NixOS lol

[wooden shore]

They have completely different purposes ???

[indigo dragon]

^

[spring cradle]

docker inside nix :D

everyone wins

[red mulch]

Yeah man what sort of idiot would use containers.

[molten oracle]

I swear this is the kind of person whose every 2nd message would be "I use arch BTW". Might replace arch with other stuff like vim.

[fading osprey]

lmao, I cannot pass comment as an Arch + NeoVim user

[molten oracle]

Maybe, but you don't seem like the kind of person whose entire personality is that they use Arch+NeoVim. Those people annoy me.

I specifically keep my dotfiles in a private repo so I don't even accidentally get mistaken for one of them.

[fading osprey]

I riff on the running gag about both, but outside of that, I don't think im too bad lol

[molten oracle]

I understand there are Arch elitists who look down on you if you even use archinstall.

But IDK how NeoVim or even Vim ended up like that

They are perfectly normal text editors.

[fading osprey]

its not about how you get there, but if you enjoy it. so long as you enjoy it, it doesnt really matter

[molten oracle]

True and Based

[acoustic pine]

Automated couple dozens of SQL entries using python kek

probably not that much quicker factoring time making the thing but

certainly less repetitive

and i can use it in the future

[molten flicker]

Are you joking?

[frozen flame]

We're all laughing at you...

[reef star]

Well I am using containers/jails, and I am also passing for an idiot, the theory has some merit...

[peak acorn]

https://xkcd.com/1319/

xkcd: Automation

'Automating' comes from the roots 'auto-' meaning 'self-', and 'mating', meaning 'screwing'.

[acoustic pine]

Haha the one benefitting from my automation the most ended up being my mates who don't need to write the code themselves

I make sure i get something in return though

[peak acorn]

Ah wellll tbf that has a lot more application than just automating something for yourself

If you are meaningfully abstracting away details so now ppl can do whatever

[red mulch]

If you have to ask this, then wow.

[hollow basalt]

this tbh

[neon oriole]

i know im not verry popular with my stance on JS as a whole. but it seems that im not allone (anymore?) with it: ps even if you swear by js, watching this might not be a bad idea , just because not using js where you dont have to might simplify and thus reduce errors and debugging the actual js you think you do need :https://www.youtube.com/watch?v=wKU65gV6FSA

[solemn solar]

As of now I kinda wanna just learn it and use it everywhere so I can learn where to not use it, if that makes any sense

[neon oriole]

somewhat but one could also say that since you start using it for everything, you might form habbits of using it also in the places where you shouldnt 🙂

the age old hammer and nail problem

[solemn solar]

I mean that was the case with me and dvh/dvw until I figured out that it, in fact, is not a perfect solution as it might seem and have since moved on to better units like rem

As of now my goal is simple: Learn javascript at its core so I can have the all three (HTML, CSS, and JS) at their core learned and have it as another thing I can do when I ultimately start a career in this
Also afaik good for learning other branches of JS and projects that make it better so that when I do move on to databases I'll have it at least a little easier

[acoustic pine]

My little newbie project with 0 JS :)

though i'm going into react so no more avoiding :(

[neon oriole]

on a completely different topic.... anyone here familiar with Qt treewidgets? i seem to have most things figured out, but for how do i remove a toplevel item in the tree, adding one is pretty simple: addTopLevelItem, but there seems to be no delete(any?)item nor a remove(any?)item. looked up and down all the methods availeble a couple times but nothing seems to fit... (and yes the screen is a qttree view of qtree view 😄 )

nvm apparenty qt speak for remove or delete is "take" :S "takeTopLevelItem"

[molten oracle]

I'm not going to watch the video because I feel like I already know what it's about. But I do agree, I see people in web dev overuse js a lot. A lot of things are much simpler by leveraging more from even html and CSS.

[molten flicker]

What, I have no social skills and it's basically impossible to convey tone across plain text? Yeah, that's hilarious /s

[neon oriole]

+i would actually say that in most cases ... like Wtf? and your using containers for That? lol , havent come across the need for them with 1 exception: a bash history manager (systemlevel so to test it its either vm or container to see if the thing works and gets set up correctly on boot, since its all sourcing and setting env vars across users from the first shell... , but other than that , needed a vm a couple times but for stuff a container couldnt do either and for all the rest a virtual env or a chroot was more than enough. possibly a sandbox to test installers but containers are overkill for that so i basicly see it like:

... everything a container is overkill for up untill chroot/sandbox] 
??? stuff a container is just right for ???
[stuff a container isnt capable off and a vm or bare metal is needed ...
```  but i wouldnt know anything that would belong on the second line

imho containers are a way of sidestepping the "works on my machine" -programmer meme

[peak acorn]

This is absolutely one of the most beneficial things

Isolating everything else in the environment for whenever you have problems is crucial

[molten flicker]

Which NixOS sidesteps much better anyway

[neon oriole]

them meme is about bad programming... but why containers? and not ship a vmdk and a vmx or an iso with syslinux and squashfs ? just to be really sure of the environment and be sure it cannot be altered to the way you know it works ?

[peak acorn]

Because deploying a docker container is way easier than booting up an iso image

[neon oriole]

that way you can be sure the user has the right kernel modules and versions , or the correct windows version with all required vcredists and environment vars set... and that the system has enough memory the right ethernet modulles , usb version ... soundcard inputs and outputs , the vm can map everything ...

[peak acorn]

Maybe youre just from a different land if thats your worry

[neon oriole]

really, booting up an iso image in a vm or with qemu is literally just running a single command or pressing the play button

[peak acorn]

For us, our builds use a known docker image base, therefore our local,CI, and production deployments are presumably as identical as possible

[neon oriole]

setting up a docker env is allot more work for the user imho

[peak acorn]

eh dunno about that but maybe

[neon oriole]

but in case the environment is not identical and that gives bugs , wont those be missed , i can see doing builds on a barebones machine if your shipping binaries anyway. or building them in a sandbox /fakeroot depending but i dont see an extra benefit of using container over a sandbox

[peak acorn]

For my work our servers are themselves docker containers

[neon oriole]

(to me it seems that a conainer has both the downides of using a sandbox , and the downsides of using a vm depending on the point of view and your requirements)

[wooden shore]

Have you ever deployed software

To others

[neon oriole]

depends on what scale

but i dont see how that would be in terms of being on the receiving end of deployed software,... most software i build (and thats a verry ffing lot) builds just fine in a sandbox ...

[molten flicker]

The only real benefit docker has over a VM is resources and (as a consequence of that) performance. NixOS has neither of those issues because everything is running on bare metal.

Containerization is not the solution

[neon oriole]

(if fact every program on my system that is available as source is build from source on my system, and i didnt require docker to build or rebuild any of them, and many have failed while building without ruining my system or without the possibility to quickly make a change and restart the build , all is done in a sandbox before being merged (on success) to the system... , other than that i did write a couple EFI drivers in the past and your kind of locked into using a vm /bare metal to test those as a container wont cut it for that either

[neon oriole]

now i wont say anything about nixos but nix ,... does not come near my system after giving it a go 1 time, and checking my /etc/passwd file after having installed a few things with it.... no thank you id rather spend the day swatting depency conflicts

[peak acorn]

irony of nixos having depdency conflicts

[molten flicker]

I think you might have been using it incorrectly 🚀

[neon oriole]

i think nixos doesnt have any for what i understand hence why i said : id rathater spend the day fixing depency conflicts (other than having them installed using the nix pkgmngr)

[wooden shore]

It sounds like none of you are the target audience for docker

[neon oriole]

as i said not nixos but the nix packagemanager... and i dunno followed the guide , but this is what my etc/passwd looked like afterwards :

[neon oriole]

fine i'll bite... what sort of software would be the target of docker. stuff like the linux kernel? KDE? gnu-coreutils?, or the complete oposite of those: domain specific applications? perhaps im wrong , but thusfar i havent come across any kind of project that has a direct comparrison/competitor that does not use docker in its dev cycle. , there is mojo/modular that doesnt seem to want to be running on anything other than ubuntu or an ubuntu environment-like container , and they go to extreme lenghts to enforce that , now i dont know why that is but i assume its not because they really think docker is better than any other way of running code and or an interpreter.

[neon oriole]

a verry similar approach to nix i think is (the pkgmanager testing distro) distri https://distr1.org/

only they use squashfs imgages and overlayfs i think

[neon oriole]

i think i forgot about one usecase where it seemd usefull indeed ,: the iruncode bot , wich needs to set up and tare down safe environments for running random usercode and returning the output without overloading , bogging down , or crashing the servers that it runs on a couple million times a day i presume... and then also stuff like godboldt but those are verry specific usecases i think, and while they might be served better using actual vm's the startup and taredown of vms is to much overhead and takes to long for the purpose i guess (https://github.com/engineer-man/piston-bot)

[red mulch]

... from scratch maybe. but most docker dev environments aren't from scratch

containers are not for "users". You don't deploy containers to an end user compute device, typically.

not to mention, support and mindshare.

If your code doesn't have to run on infrastructure made and/or supported by anyone else, and doesn't have to be part of any established CI/CD or build system and doesn't have to have any sort of wide market support to ensure ongoing support, sure, do whatever the fuck you want

and the shit about using a VM instead? holy shit the overhead in a VM is so huge comparatively. And for almost everything, almost all of the time, it adds nothing.

nix and nix style deployment might well be the future, but it doesn't make containers bad.

[keen sorrel]

A dockerfile provides the ability to just instantly have the entire development toolchain of some project you've never interacted with before. Then being able to instantly remove that entire toolchain when you're done with the project. It saves a hell of a lot of effort and keeps your system clean when exploring oss projects

[wooden shore]

Everything @red mulch said 👍

[red mulch]

it's almost like I manage code and infrastructure professionally and have done for decades

[wooden shore]

Containers means you don’t need to care about how your host is configured. Makes it easy to spin up the same consistent environment, which I guess nix does too but then you get a not-so-great ux, I can install docker on any os and my software will always run the same

I also have all my devs machines set up with docker so they can replicate the deployment targets environment. it’s easier to troubleshoot issues when your environments are all setup the same

[frozen flame]

there are certainly places where containers arent ideal, and can be more of a burden than just farting a binary onto a server. but that place isnt corporate prod

[neon oriole]

yeah ture but i gave one use-case that i missed , fast and safe arbitrary user code execution... , other than that for efi/kernel/driver stuff a container wont do so baremetal or vm required. and for anything mentioned so far , not in the specific use-case i just mentioned a sandbox build will do (maybe im missing more, im open to it but so far nobody has mentioned a case where a container is needed over a sandbox/chroot) even if someone else does support for your software. and yes a vm is allot more overhead but not that much in relation to a container , vs a container and a sandbox/chroot. , take thinapp(i think is the name, its dead now but the priciple stands)... its technically running an app in a vm. but the overhead is not that crazy. compared to containers. since unless your emulating everything hardware is almost transparant to the vm.

the fact that something is enterprise doesnt mean its any good or done efficiently , the opposite usually (https://www.youtube.com/watch?v=FyCYva9DhsI)

(maybe i found another usecase , code benchmarking, when theres allot of different compilers an languages to be tested each run , i assume that in that case containers might be a good solution to so have an identical starting point for each, like https://plummerssoftwarellc.github.io/PrimeView/)

[molten flicker]

Yep. That's normal. You were reading the install script's output, right?

[neon oriole]

cant remember , is over a year ago 🙂 but i think i just did the recommend things , and i know there was a mention of having to create a user for nix (i assumed similar to portage user or apache, smbuser, to run the daemon as so it doesnt have to run as root... but didnt expect it would create 500 extra new users on its own without any reference to what they are for or what they are running, (eg if i would have installed firefox trough nix , i could live with a nix-firefox-#uuid user} but a nix#guesswhat , i dont like, + i was only testing it to see if it had more promise than snap or flatpaks so ...

[molten flicker]

32, not 500. Which is the typical upper limit for CPU cores nowadays.

I have no clue what the technical reasoning behind it is though.

[neon oriole]

well not 500 but more than 32 :p

[molten flicker]

maybe its the number of cores your system has.

[molten flicker]

At any rate, that's totally normal. You have to read what the installer is telling you.

It explicitly says it will do this and it needs to use sudo, are you okay with this?

[neon oriole]

yeah then i wouldnt mind since that would mean only 4 entries ...

its also possible i installed it with my sistems package manager btw ...

[molten flicker]

it shouldn't be, at least to my knowledge

[acoustic pine]

Welp

Seems like my CSS is slowly devolving into a spaghetti again
tried my best keeping it organized

but the bandaid fixes started to pile up

It was working so well but someway down the line i changed something that broke all the relative height functions

meaning i had to resort to annoying absolute pixel value instead of easy percent values

[solemn solar]

Interesting, I tend to keep mine minimal and I hardly have issues with it

[acoustic pine]

I probably need to rewrite a big chunk of the CSS and have the modularity fixed

I think it's because i broke something on one of the parents height settings

[peak acorn]

just use tailwind n call it a day 🙂

[solemn solar]

nuh uh

[peak acorn]

absoooo lutely

[solemn solar]

I refuse to use anything other than basic HTML and CSS for learning something else would require effort that I do not have
(Also it'd involve regaining my muscle memory from scratch which isn't ideal)

[acoustic pine]

Maybe i'll look into it
but for now i like having deeper knowledge with base CSS

Need to use flexboxes more properly

[wooden shore]

You should definitely learn vanilla css before learning tailwind

[solemn solar]

And that's why I don't wanna learn Tailwind for now

[vestal crest]

so, i used to use spy++ if i wanted to get the hwnd or hclass of a particular window in windows, would also let me look at what the various classes and names and such were for the text boxes and buttons within that window too... spy++ is functionally useless on windows made in dotnet... does anybody know if any tools i can use to do a similar amount of spying to find window information for dotnet (xaml) made windows

[molten oracle]

Using tailwind is like writing CSS using classes instead of just writing CSS. And it adds bloat. Unless it's a big project with a lot of repeating classes that cannot be divided into components, it's pretty pointless to use it imo.

[wicked scarab]

You can always use VSCode extensions to hide the class names, agree though it gets way cluttered fast

As someone decently new to dev land, it’s always been kinda nice to have access to the element directly too rather than having to go to global to find it and name it. Although Global has always nice for themes like fonts or buttons, and even for formatting flex vs grid depending on the page

[keen sorrel]

https://github.com/icsharpcode/ILSpy
https://www.jetbrains.com/decompiler/
Both ILSpy and dotPeek feature the ability to decompile baml back into xaml.

[vestal crest]

yeah, already decompiling, was hoping there was a simple tool like spy++, but doesnt seem like there is. ty 🙂

[arctic lantern]

got some rgb lights chrismass
but they need an app that needs a login (no chance of that and it dont even work anyway)
so been RE them for few months and not i gotten them working with a random python script
next step homeassistent intgration lol

[gilded stream]

I know the basics of c++ and can program stuff too. I wanted to know what kind of projects can i do to get a job in the field of game development with c++, and where can i get started. I AM NOT A BIGINNER IN THE FIELD OF PROGRAMMING BUT I AM IN THE FIELD OF GAME DEVELOPMENT SOOOOO

I need advice

[vague parrot]

i fucking love react, i don't have enough parenthesis tho

[solemn solar]

Yeah no you need more parentheses, I'd at least double or triple that

[nocturne galleon]

Add more newlines 😭

[vague parrot]

No, spaghetti Code It is

[peak acorn]

make games, simple 🙂

game jams, larger projects, etc

[spark temple]

Please use prettier and eslint. My eyes are bleeding.

[vague parrot]

@gilded stream I usually have fun making a simple 2d game

A simple "shooter", enemy spawn from the side, your character shoot when you don't move etc etc

[gilded stream]

but if i want to work for a company who make 3d games then

[vague parrot]

You don't start by programming YouTube my dude, start with the basics

3d IS just 2d with extra steps

[gilded stream]

i have done some shit with 2d

so like 3d for example game engines etc

you have a roadmap for it?

[vague parrot]

I'm more audio dev than full ass dev, so most of the time I don't create the game from the ground up

I just try interactive audio and shit

[regal folio]

for (int i = 0; i < LOAN_BOOKS; i++)
{
    if (2>3);
    {
        printf("\t*%s, By: %s, Retrun Date:", member->LoanBooks[i].Name, member->LoanBooks[i].AuthorName);
        print_date(&member->LoanBooks[i].ReturnDate);
        printf("\n");
    }
}

for some reason the code inside the if statement runs no matter what.. i tried 2 > 3 to see if i was right.. it still runs like how?
pls help

[fading osprey]

could it be the semicolon after the if condition?

[regal folio]

fml T_T

[wooden shore]

This is called a virus

[silk eagle]

well python has ImportError for a reason

[fossil flame]

pain

[vague parrot]

at least you don't have to deal with function.then() giving you half cooked object

it's 50/50 a promise and an object, i can't deal with js

[regal folio]

void add_member(LibMember* members, int nMems)
{
    LibMember* temp = (LibMember*)realloc(members, (nMems + 1) * sizeof(LibMember));
    if (temp != NULL) members = temp;
    members[nMems].Name = (char*)malloc(100 * sizeof(char));
```whats the problem here?
```c
void main()
{
    int nMems = 0;
    int loop = 1;

    LibMember* arr = (LibMember*)malloc(1 * sizeof(LibMember));

    while (loop)
    {
        switch (menu())
        {
        case(1):
            add_member(arr, nMems);
            nMems++;
            break;
``` this is the main

i feel like im not working with the pointer of members in a good way..

[peak acorn]

Is there an error?

Whenever you do case 1, reallocate members to be one bigger, and allocate memory for their name. Seems fine
(although if realloc fails i think youll crash)

[frozen flame]

Babe wake up new RCE dropped

https://news.ycombinator.com/item?id=39865810

rkta

Backdoor in upstream xz/liblzma leading to SSH server compromise

Hacker News · Mar 29, 2024, 4:16 PM

[fossil flame]

nope no thankyou how is that even possible

im writng a windows app that turns scripts into apps like just normal python scripts and then making a ui editor simaler to when you make a website but better

[frozen flame]

Have you considered nuitka

And maybe a wrapper around pyside

[solemn solar]

I'm lost at this point. I want to make it so that pressing on the doordiv also closes the navbar, and that part works when logging using the mobileconsole log I've set up, but for some reason it just REFUSES to work and actually remove the navbar. It triggers the closing state, but doesn't move the navbar off and if I press anywhere that isn't the navbar it does close

[regal folio]

    printf("Enter Book Name: ");
    scanf("%s", bookName);
    (*members)[search_id(*members, nMems, idInput)].LoanBooks[emptyIndex].Name = bookName;

    printf("Enter Author Name: ");
    scanf(" %s", authorName);
    members[search_id(*members, nMems, idInput)]->LoanBooks[emptyIndex].AuthorName = authorName;

    // Get the current time
    time_t now;
    struct tm* local_time;
    now = time(NULL);
    local_time = localtime(&now);

    // Extract today's date
    int day = local_time->tm_mday;
    int month = local_time->tm_mon + 1;  // Months are 0-based, so add 1
    int year = local_time->tm_year + 1900;  // Years are counted from 1900
    
    month++; // Adding 1 month from today for the return date

    if (month == 13) // Adding a year and changing the month to 1 if it's the last month of the year
    {
        month -= 12;
        year++;
    }

    members[search_id(*members, nMems, idInput)]->LoanBooks[emptyIndex].ReturnDate.Day = day;
    members[search_id(*members, nMems, idInput)]->LoanBooks[emptyIndex].ReturnDate.Month = month;
    members[search_id(*members, nMems, idInput)]->LoanBooks[emptyIndex].ReturnDate.Year = year;

    members[search_id(*members, nMems, idInput)]->nBooks++; // Adding 1 to the number of books borrowed

    printf("Book Loaned Successfully!\n");
}``` can someone please tell me why the date gets transferred by reference  but the book name and author name wont get transferred?

[sudden cosmos]

cause authorName and bookName are just pointers to local variables not values like the date. you should copy them with strncpy instead of reassigning the pointer.

something like

int member_id = search_id(*members, nMems, idInput);
if(member_id<0) return -1;
LoanBook_t* loaned_book = &members[member_id]->LoanBooks[emptyIndex];

scanf("%s", bookName);
strncpy(loaned_book->Name, bookName, sizeof(loaned_book->Name));
scanf(" %s", authorName);
strncpy(loaned_book->AuthorName, authorName, sizeof(loaned_book->AuthorName));

assuming Name/AuthorName are arrays and not just a pointer. if they're pointers you'd have to get the destination length a different way.

[nocturne galleon]

Hi everyone, this is possibly a very stupid, very noob question, but would react help with making my website look better style/layout wise? or is it purely for interactivity?

for example, the react docs site, is the nice clean look achievable because they use react? or is it just styling via css?

[red mulch]

Omfg this is burning my eyes. Why would you not use a date library

[fading osprey]

its entirely down to the way that it is styled with CSS, React (and any framework for that matter) is entirely there to allow modularity and a higher degree of interactivity with much more ease than vanilla JS

[regal folio]

i sorted it by using a function that returns a struct of a date now

[red mulch]

Also storing the date as seperate y/m/d, it's horrific

[regal folio]

tell that to my professor

[red mulch]

Uuuuugh. I am physically affected by it! 😄

[regal folio]

whats wrong with it though, i have a struct with y/m/d ints in it

int day = todays_date().Day;
int month = todays_date().Month;
int year = todays_date().Year;

month++; ; // Adding 1 month from today for the return date

if (month == 13) // Adding a year and changing the month to 1 if it's the last month of the year
{
    month = 1;
    year++;
}

members[search_id(members, nMems, idInput)].LoanBooks[emptyIndex].ReturnDate = (Date){ day, month, year };``` this is the same code after sorting it

[red mulch]

Im sure time delta libraries exist for C, so you would define the time delta as 1 month and then use the time library functions to increment it

[regal folio]

they specified that i should use <time.h> library

[nocturne galleon]

aah ok, that has made it clearer, thank you.

One othe question, if I just want to use react for the sake of having JSX, how would I add it to my otherwise vanilla site? I have a fleshed out API already so I don't need one of the react frameworks I see on their site that is fullstack

[fading osprey]

I'm afraid I cannot answer that as I don't use react myself, I am more of a svelte guy, but im sure someone else might be able to help

[red mulch]

and then you could define the loan_period as a delta using a const, which makes it much easier to read AND change

[regal folio]

my code doesnt have problems so i wont touch it lmao thank you though

[red mulch]

haha

Fair enough. I mean it's functional, that's what's required.
It's not like anyone is going to using it.

[regal folio]

i tried searching everywhere and tried everything, thats the closest i got to having the scanf only reading 1 charcter without looping for no reason if the input is longer

my question is , is there a more efficient way to do this? since in my method i used 500 bytes, if there is a better way please tell me

and if i type more than 500 characters it will crash

[hollow basalt]

then don't

[regal folio]

my professor could....

[hollow basalt]

you could

[tight quail]

does anyone here use JetBrains Webstorm? is it worth using over VSCode+extensions?

[nocturne galleon]

hm, this is a really weird issue but maybe someone here knows, i'm implementing microsoft's oauth interactive flow with a javafx webview, and it mostly works, but there's a weird issue where i need to put the right password twice (as in, the first time it will always say it's wrong) for it to work

this is reproducible across multiple attempts and multiple accounts

which i find extremely weird since all of that should be getting handled on microsoft's side, all i do is display the page and listen for a redirect

[peak acorn]

avoid crashes by always using string N functions

which will handle when strings are too large

And then i think on scanf you can do %500s to limit the lenght

[neon oriole]

or the really inneffective approach using getchar : ```c
static char *readPipe(){
unsigned int sR = 0 ,
sA = 8 ;
char cI = '\0';
char *pipe = malloc( sA );
char *str;
char *pPipe;

// sPipe = getline(&pipe, &len, stdin);
while ( (cI = getchar()) != EOF ){
    //if char counter read == allocated counter(-1 for adding \0) -> rallocate double the size
    if (sR==(sA-1)){
        sA=2*sA

; pPipe = realloc(pipe,sA);
if ( pPipe == NULL ){
str=NULL;
goto END;
}
pipe = pPipe;
}//if
pipe[sR++] = cI;
}//while
pipe[sR]='\0';
str = (char*)malloc( strlen(pipe)+1 );
strcpy(str, pipe);
free(pipe);
END:
return str;
}//readPipe

[peak acorn]

#include <stdio.h>
#include <stdlib.h>

#define MAX_LENGTH 4
#define PROMPT "%4s"

int main() {
    char* inputBuf = malloc((MAX_LENGTH+1) * sizeof(char));
    printf("DB System Menu\n");
    printf("\t1. A\n\t2. B\n\t3. C\n");
    printf(": ");
    scanf(PROMPT, inputBuf);
    printf("You typed %s\n", inputBuf); // Inputbuf will be truncated after the 4th character. IE: "ABCDE" will print "ABCD"

    return 0;
}

[peak acorn]

You migh wanna look too i dunno

[nocturne galleon]

shouldn't the MAX_LENGTH+1 be in parentheses due to operator precedence?

[vestal crest]

so, just found this infographic on the XZ thing. explains quite well how complex, and "long game" the whole thing was: https://twitter.com/fr0gger_/status/1774342248437813525/photo/1

Thomas Roccia 🤘 (@fr0gger_) on X

🤯 The level of sophistication of the XZ attack is very impressive! I tried to make sense of the analysis in a single page (which was quite complicated)!

I hope it helps to make sense of the information out there. Please treat the information "as is" while the analysis…

Twitter · Mar 31, 2024, 7:45 AM

[neon oriole]

does it have to be scanf ? as i posted in my example i use getchar() to read from stdin char by char... and i allocate only a verry small piece of memory to begin with. then loop the getchar() if i am running up against the limit of what i allocated before i get an EOF on stdin, i allocate double of what was allocated with realloc... (to not have to realloc wich is quitte expensive each loop , it might end up overallocating allot with verry long strings but never more than double of what was needed so ... i think that tradeoff is worth it

[peak acorn]

uhhhhh probably yeas lmao, will edit incase anyone else looks at it

[peak acorn]

doesnt have to be scanf of course, but its way simpler

[neon oriole]

this poc is also in the way to over my head category :

its a bout the vuln iun nftables so different issue but also pretty recent

[vestal crest]

its insanely complex, and this "one person", is almost certainly multiple people... and personally, i think what it shows up more than anything is... autotools is a really bad idea, because without autotools and it obfucating away the build process, this would have been spotted immediately

[neon oriole]

my was about a vuln in the kernel ,but as new as the xz thing, only what i posted is even more convoluted but was found and posted as a poc by a university /research group so not malicious unlike the xz story

[vestal crest]

another thought i had was... when you are allowed to find something bad, its usually to distract you from finding something worse... cartels do it with the police all the time... "we found 1000kg of drugs" "yes, but you MISSED the 100,000kg of drugs that went through the other port"

[neon oriole]

also the stimplest would be to use getline on stdin, only works if there are no newlines expected in the possible input ofc:```c
char *line = NULL;
size_t len = 0;
ssize_t lineSize = 0;
lineSize = getline(&line, &len, stdin);
printf("%s (%zu chars)\n", line, lineSize - 1);
free(line);

[neon oriole]

nah they send mules with one 1k in there backpack trough airports for that purpose... the people who think thats legit and the customs are payed of and they will get a nice payday for smuggling a kilo , when there are literal tonns being smuggled in containers must be some of the most naive people on the planet

[neon oriole]

the port scanners can only scan 1% of suspect traffic (of traffic intended to be imported ) and they cannot do it in any timely fashion so the cartels just send double, if it gets trough nice if it does not , thats on average maybe 1/100 containers so they make the 99 that get trough cost 1% more to the buyers

much cheaper than paying off scan and customs officers ...

or offering them stuff to be caught..

this port is 15 minutes from where i live :p https://www.youtube.com/watch?v=EvWWuhBFTpU

[regal folio]

void add_member(LibMember** members, int nMems)
{
    if (nMems == 300)
    {
        printf("ERROR! Members max capacity reached 300!, please delete a member to add another.\n");
        return;
    }

    // Increase the LibMember array memory size by the size of LibMember
    LibMember* temp = (LibMember*)realloc(*members, (nMems + 1) * sizeof(LibMember)); 
    if (temp != NULL) *members = temp;

    (*members)[nMems].Name = (char*)malloc(NAME_SIZE * sizeof(char)); // Dynamically allocate memory for the member's Name

    while (1) // Loops until correct Name inputed and adds it to the new member
    {
        int counter = 0;
        printf("Enter New Member Name: ");
        scanf(" %[^\n]%*c", (*members)[nMems].Name);
``` how can i make it so the name gets saved in the pointer and not a local value?

[peak acorn]

uhh

[vague parrot]

why do everything that i do look so cursed ?

useFocusEffect(
    React.useCallback(() => {
     /* callback Function and shtuff*/
    }, [,refreshing])
  );

[peak acorn]

why do that?

[cloud knot]

that block of code hurts. 2024/01/31 +1 month in your case creates 2024/02/31 😄
also 2024/03/31 + 1 month makes it 2024/04/31. Which also hurts.

[cloud knot]

React in general looks cursed

[vague parrot]

it's the dependancy array of the useCallback function (basically, the function will be used if something in that array changes) , if you leave it empty, it will use th function on component mount, so i tried (and it worked)
[] ==> function used on mount
[var] ==> function used on var modification
[] + [var] ==> [,var] ==> function used on mount, and on var modification

it's cursed, but it works

[regal folio]

They told me to make it simple and assume 30 days in a month

[cerulean path]

seems like a weird assumption to me but if it works. it works

[quick agate]

Pascal or TCL for a school assignment

[empty pier]

This is how you centre a div, right?

[keen sorrel]

No, you should be using strings instead. Definitely more efficient than integers

[acoustic pine]

What kind of cursed shit is this

[empty pier]

🤣

[solemn solar]

dear god why must you curse me on this fine day

[cloud knot]

reminds me of the classic 😄

private boolean isTrue(boolean val) {
  return val == true;
}

[bitter mantle]

hmm guys can someone check some sql, php and css code to see if its working and all

cause idk how to open a database or a server in mssql extension

in vs code . so yeah can someone check it and tell if the code works properly ?

ill send the code here if you want .

[tardy barn]

Writing a dockerfile, can't manage to copy this one file. It errors out as "/run_bot.sh": not found
" Ive double checked spelling, tried changing it to ADD and tried specifying the WORKDIR. Any thoughts? There is no .dockerignore.

COPY ./run_bot.sh /app/run_bot.sh

[peak acorn]

I mean /run_bot.sh is an absolute directory from root, so that's not /app/run_bot.sh

[molten flicker]

Can't wait for (technical) feature-parity with W*ndows, though. Imagine, the NT kernel as it should have been. It will revolutionize gaming even more than Atlas already has.

Oh, React... not ReactOS.

Yeah, React is cursed.

[neon oriole]

javascript is a blight to humanity so ... :p

[static oar]

Does anyone here program in Erlang? Curious what if any types of projects people might be actively using it for.

[molten oracle]

Elixir, which builds on top of erlang is what discord's backend is written in.

As for direct uses, I can only think of RabbitMQ and WhatsApp on the top of my head.

[indigo dragon]

anybody here know RISC-V nuts and bolts?

reading through the spec and not sure what C2 means here

https://drive.google.com/file/d/1s0lZxUZaa7eV_O0_WsZzaurFLLww7ou5/view

I'm guessing that's just a literal that's in the field since none of them are larger than 3 and the C suffix is there because it's a compressed encoding

[sudden cosmos]

the raw bits for the different opcodes are in the tables after that, 16.5-16.7. but yeah C2 is b10. 00 = C0, 01 = C1, 10 = C2, 11 = 32-bit instruction not a compressed one

[indigo dragon]

Yea that makes sense

[dreamy lance]

Can anyone help me in deploying a Node Express, MySQL, React app? I have my application ready but I have no experience in deployment of any sort

[slender brook]

i might help

[dreamy lance]

ill send a dm bro

[frozen flame]

Deploying to what

Deploying to a vps is different from deploying to a managed site

[molten oracle]

I think they're asking for what they should do as a newbie with no knowledge of deployment. And I'd personally recommend a managed site first.

I'm not sure about MySQL but railway is nice for free node app hosting.

[undone glen]

I was wondering if i can get some help with something RQ. I have done some research and asked some people but i have not reached an answer. I am developing a car game in unity and while the car drives as expected. the wheels do not render when playing the game. (They do in the editor) I have no idea why this could be. The wheels are a child of the main car body.

[untold badge]

Is there any use for an NPU as a dev yet?

[keen sorrel]

Ofc. It's pretty unusual when devs aren't the first to be able to play with new toys technology.

https://www.amd.com/en/developer/resources/ryzen-ai-software.html

ML/AI devs can already get onnx models deployed to it with just a few steps

[untold badge]

I moreso meant as a normal dev, not a super duper early adopter

Never used onnx tbh

[keen sorrel]

What? This is for normal devs. This is all standard dev processes for AI/ML.

[untold badge]

I'm blind, missed where it mentioned PyTorch and Tensorflow

I'm guessing you can't train with them yet?

[keen sorrel]

Afaik the AIE "cores" in the NPU only support 4-bit and 8-bit ops. You wouldn't really be able to train effectively, if they even have enough TOPs to do so fast enough to compete with a GPU

[untold badge]

To you forsee us being able to run LLMs or coding assistance tools any time soon on them?

My use case is stuff like Copilot (coding), Stable Diffusion (fun), and maybe image recognition (some random python libraries) instead of developing AI tools lol

[keen sorrel]

I would bet LLM models are available today on hugging face that will run on them. Coding assistance tools will probably take a little bit more effort because they involve more software than just running the ai model itself.

https://huggingface.co/models?other=RyzenAI

These are the models that are ready to run with a single click. Other models will require quantization and deployment steps to use. Whether a model survives quantizing and deployment to the NPU with acceptable accuracy is another question though

[empty pier]

what does your development toolbox look like?

i also dont develop in roblox nor do i use SQL but roblox studio came with roblox and i have SQL because i was testing something

[tawdry otter]

my toolbox looks like this:

• Intelij
• PHPstorm
• MySQL Workbench
• plain ol linux terminal
• hetzner

[fading osprey]

My toolbox looks like this:

• Alacritty
• neovim

[solemn solar]

My toolbox is very simple
VSCode and Firefox Developer Nightly. Sometimes Arc (My daily driver browser) for testing chromium stuff or vertical tab users, but that's all

[raw plume]

My toolbox is

• PHPStorm
• Laragon
• WinSCP
• Contabo

[neon oriole]

0no debugger/test runtime ? That is some balls of steel lol code the whole thing begin to end and have it work just like that

[fading osprey]

Never said my code works, just that that's my tool box

[keen sorrel]

I might be forgetting something I just haven't used in a bit but this is mine rn:

• VSC
• Rust toolchain
• Vivado+Vitis+DocNav
• Powershell
• Vulkan SDK
• Sysinternals (windows only)
• WSL2 (windows only)
• Radare2

[molten flicker]

Kitty, Neovim, NixOS btw, gitui, bacon, tmux (rarely tho)

Recently swapped bash out for nushell but begrudgingly so (I still drop into bash from time to time)

[grizzled steeple]

• WebStorm/IntelliJ
• NVM
• a Terminal (usually the Windows or Mac Terminal, depending on the system I work on)
• Firefox & Brave (I prefer Brave over standard Chrome for testing Chromium)
  I don't need many tools as a web dev I guess xD

[molten oracle]

Mine change drastically based on how I might be feeling that day

Could be a rust nvim day, could be a C vscode day, could be a TS/Webstorm day, could be a PHP notepad day.

Languages and editors/ide are paired at random as well

[wooden shore]

Wtf

[crude belfry]

P

[dreamy lance]

SELECT
ROW_NUMBER() OVER (ORDER BY datastructures.DSBatch DESC, actionresults.ActionSet DESC) AS RowNumber,
datastructures.,
actionresults.
FROM
datastructures
LEFT JOIN
actionresults ON actionresults.DSID = datastructures.DSID
WHERE
datastructures.AccountID = ?
GROUP BY
datastructures.DSBatch,
actionresults.ActionDate
ORDER BY
datastructures.DSBatch DESC,
actionresults.ActionSet DESC;

any idea why this query behaves differently when used on ubuntu mysql? was working on it with xampp mysql on development and it works just fine. the query in ubuntu mysql seems to retrieve more records than expected. might have to do something about GROUP BY idk

Response from ubuntu mysql with the same query looks like this:

[
{
"RowNumber": 1,
"DSID": 432,
"AccountID": 94,
"DSBatch": 1,
"DSName": "Frequency List V2",
"Threaded": 0,
"Frequency": 100,
"Capacity": null,
"Type": "CUSTOM LIST",
"UserAddedPivot": 0,
"DateCreated": "2024-04-12T08:27:42.000Z",
"ActionNumber": 594,
"ActionSet": 1,
"ActionType": "Add",
"StartingIndex": 0,
"EndIndex_Count": 5,
"Direction": "Add Start",
"SpeedMS": 0,
"SpeedNotation": "O(1)",
"Size": 5,
"SizePointers": 2,
"PointersAdded": 2,
"SizeAdded": 5,
"SpaceOccupied": 130,
"SpaceNotation": "O(1)",
"SpaceAdded": 100,
"ThreadsUsed": 0
{
"RowNumber": 2,
"DSID": 435,
"AccountID": 94,
"DSBatch": 1,
"DSName": "Dynamic Array",
"Threaded": 0,
"Frequency": null,
"Capacity": 100,
"Type": "TRADITIONAL ARRAY",
"UserAddedPivot": 0,
"DateCreated": "2024-04-12T08:27:42.000Z",
"ActionNumber": 597,
"ActionSet": 1,
"ActionType": "Add",
"StartingIndex": 0,
"EndIndex_Count": 5,
"Direction": "Add Start",
"SpeedMS": 0.1,
"SpeedNotation": "O(1)",
"Size": 5,
"SizePointers": 0,
"PointersAdded": 0,
"SizeAdded": 0,
"SpaceOccupied": 426,
"SpaceNotation": "O(1)",
"SpaceAdded": 10,
"ThreadsUsed": 0
]

while the right behaviour should be like this (from xampp mysql):

[
{
"RowNumber": 1,
"DSID": 440,
"AccountID": 96,
"DSBatch": 1,
"DSName": "Frequency List V2",
"Threaded": 0,
"Frequency": 100,
"Capacity": null,
"Type": "CUSTOM LIST",
"UserAddedPivot": 0,
"DateCreated": "2024-04-12T08:28:30.000Z",
"ActionNumber": 618,
"ActionSet": 1,
"ActionType": "Add",
"StartingIndex": 0,
"EndIndex_Count": 5,
"Direction": "Add Start",
"SpeedMS": 0,
"SpeedNotation": "O(1)",
"Size": 5,
"SizePointers": 2,
"PointersAdded": 2,
"SizeAdded": 5,
"SpaceOccupied": 130,
"SpaceNotation": "O(1)",
"SpaceAdded": 100,
"ThreadsUsed": 0
]

[cursive crater]

Might be because one uses Mysql and the other is on MariaDB or different server versions? I cant do a deep check because I'm not at home currently

[dreamy lance]

yes, apparently mariadb works a little different when it comes to GROUP BY method. I had to rewrite my queries a little. Alls well now. Appreciate ur response bro

[cursive crater]

Later Mysql versions adhete a bit better to the official SQL standard so itd good you rewrote it. Should prevent some future headaches 😊

[frozen flame]

Using almost anything other than MySQL would remove a lot of headaches tbh

[acoustic pine]

yep github fugged and i can't log in now

smh

owned by microsoft the 3 trillion dollar company and they still fuck up email service

[solemn solar]

I've had problems like this on quite a few services, hopefully gonna be resolved soon

[acoustic pine]

time to ditch shitty email based verification

sooo many fucking times i have trouble logging in in so many services thanks to email not coming in

[solemn solar]

(Not today problems, just over the course of my life I've had them)
Also yeah, just go set up 2FA, way safer regardless, plus it works even if your phone is offline

[acoustic pine]

it's not 2FA

it's device verification since i reinstalled windows and my home internet Dynamic ip changed

and i don't have mobile github installed so horray

[solemn solar]

I see, weird that it requires an email OTP, never had that requirement

[acoustic pine]

and today i actually got motivation to finish my personal project

and github fucked my day

[solemn solar]

That sucks, hope it gets fixed soon though

[wooden shore]

https://customkeys.gg/builder/ I am creating a website where you can design a custom keyboard online with a 3d editor

[nocturne galleon]

nice

[cloud knot]

i guess i found a project to do for a feature i can't seem to find anywhere - XML NFO & image export from Plex. Yes, one can do it via Tautulli, but in a format i don't like (i prefer my media be in "separate folders, movie.nfo and images in those folders etc" format). There were such functions for older Plex when it had plugins. I guess a little fun project, the really big question will be to pick the right UI/UX stuff for it to be universal. Let's see how long will it take me to do it😄

[keen sorrel]

It only exists when you lack 2FA.
https://github.blog/changelog/2019-07-01-verified-devices/

[sick flame]

anyone know java enough to help me with an error im getting?

[meager jasper]

@sick flame depends on the error i guess

[limpid reef]

What @meager jasper said - I don't want to be that guy but also... this is kind of like asking "hey, anyone here know how car engines work?" because you want help with very obscure problem that may or may not have anything to do with car engines at all. Some programming related questions don't even need one to be familiar with programming language X because the same logic might apply to all languages. Put more simply, just ask your actual question then wait to see if someone is able to help answer it. 🙂

[meager jasper]

I use Java in my dayjob and have done so for 5+ yrs so go ahead and ask and I'll see what I can do

(also, unpopular opinion: I like Java and it is a good programming language overall)

[spiral tulip]

hi, i'm useing gml for a game i'm making, i'm useing this script for collitions problem is when i walk into a wall it works half the time. other times theres a invisable barrier and i can walk through certen walls.

if place_meeting( x + xspd, y, obj_wall ) == true { xspd = 0; } if place_meeting( y + yspd, x, obj_wall ) == true { yspd = 0; }

[nocturne galleon]

my toolbox looks like this:

• Visual Studio
• VS Code
• MySQL
• python
• nodejs

[half dome]

Whats in place_meeting()? Also, how do you move the Player, just like in the check with x + xspd and y + yspd in every cycle it is checked?

[cloud knot]

also are you sure you want to swap order of arguments ? first call does x,y; while second does y,x

[spiral tulip]

Sorry guys some guy helped me yesterday

[golden mural]

I have my old workstation I upgraded from a Ryzen 1800x to a 3900x is this thing capable enough for use as a workstation for data science development

[topaz flare]

Depends on what you’re doing

[golden mural]

well i origanlly bought the machine for KVM gpu pass through becuase i Run linux and i wanted to develop on linux run the software i write on windows

[elder robin]

i need some suggestions on what features i could do for my game project
ive went with making blackjack and want to include some other things if anyones got any ideas tell me pls...
ty

[honest juniper]

Hello HTML doubt, suppose I'm using image map, how am I supposed to find out the coord of the part of image I want to map

[wooden shore]

Map to what

[neon oriole]

I use wine for that 🙂 with the idea of ,if it works on wine itl prolly wirk on windows just fine. (Only in C ive found that the max stack size isnt always consistent with the windows stack size of 1MB *linux gives you 10MB, and im sure I went over the 1 mb in some version of wine and it still worked but thats a ling time ago so) ps i use the zigg cc compiler on linux to compile the windows binaries aswel zigg cc is a drop in replacement for gcc. And can compile for any platform,os, enduanness, ...) you could want

[golden mural]

why are you using the zig compiler instead ?

[peak acorn]

p sure its better lole

[neon oriole]

because it cand do this (on windows , on linux and on macos, and freebsd ,... )

[golden mural]

the image came through rather blury - i cnanot see anything

[neon oriole]

try opening it in another tab 😛

and then you can see it good quality everytwhere normall

or whats it open in browser link at the bottom of it

if it still fials i'll reup another gif of the same 🙂

targets zig cc can build for :http://0x0.st/XoHZ.txt

[golden mural]

so its just one build system for N targets

I usally forgo a make file - for the life of me the syntax is impossible to get the hangof - i end up just using a shell script with arguments for debug or optimize - which i know is kinda cursed and not portable

[neon oriole]

yes but zig is actuallty a language, but their compiler includes a drop in c and c++ compiler because zig can use c libraries and integrate with them

[neon oriole]

its almost the same as a shell script actually , i only do the verry basics in it actually sec ill show whats in that one

[golden mural]

I havent found a - good build system for C++ and C

[golden mural]

i write for all sorts of targets Dos, windows, Unix - im a retro hobbist but like the *nix enviroment for development

[neon oriole]

real old 16bit ms dos i dont think zig cc can compile for

if you use Cosmopolitan ligc (https://justine.lol/cosmopolitan/index.html) you dont even have to careate 2 seperate binaries for linux and windows, it crates 1 binary that can run on both :p

[empty pier]

I swear this happens every time I show Codey some of my code 🤣

[leaden sorrel]

Trying to get back into freelance website development, what are some good ways to advertise yourself and find clients? It's been a long while and a lot of mine were through word of mouth

[empty pier]

Most important thing is make sure your website is good. Make sure it's clean and responsive and dedicate a page showing some other websites youve made in the past.

Make up some random companies and make fake pages and host them on subdomains if you have to, good portfolio website is the main thing. After that social media is your best bet if you don't have much of a budget. Use Search terms like "front/backend developer needed" and sort by newest and then link your website there

And discord servers are a good place too, there are some with channels for putting up ads and there are also channels with clients looking for devs

[leaden sorrel]

Thank you @empty pier

[modest tartan]

what language should i learn i want to start with one that can do a bit of everything

[leaden sorrel]

A bit of everything? Python

[modest tartan]

What’s the best way to learn python

[leaden sorrel]

depends on how you learn i guess

read the documentation, watch some tutorials, dissect other people's code

[modest tartan]

I just need someone there to keep me focused on learning

[spark temple]

A lot of programming is learning on the fly. Find out how you digest big concepts and then how you solve for specifics when implementing.

Sorting information by type and problem space will help you be more effective.

[acoustic pine]

WSL2 or Hyper V

[solemn solar]

Oh boy VSCode is now copying my habit of coding CSS in HTML

[wooden shore]

The one they recommend

[empty pier]

python sucks, use javascript /hj

i mean js sucks too but its super versitle

c# is really good but its harder to use for web development

javascript is a good all rounder, it can do desktop apps, mobile apps, console apps, web apps, web frontend, web backend

it just tends to have a mind of its own and a lot of things it does dont make any sense

[modest tartan]

javascript is not meant for most of the it purpose was for web front end

[simple canopy]

hahaha I have seen this before... and sometimes there is a point to writing CSS in HTML files (e.g. for emails) or very very simple webpages

[simple canopy]

yeah I do agree. JS sucks too, but so does basically every language in some way or another. And yes, there's the "Moore's law of JS" that essentially says "What can be done in JavaScript, will be done in JavaScript"... And it's not that great at things that aren't front end, but it's still so much faster than Python in most task (and yes I know, there are Python libraries, that perform really really well, but plain python code is very slow very often).... And JS's Syntax is not too far from other major programming languages like the C family and Java, unlike Python's syntax, which is very distant from everything else I know. But I would still argue, learning Python as (one of) your first programming language(s) is a good idea, as it forces you to indent and you learn to make your code legible as a result of that

[modest tartan]

well it telling you that wsl2 is recomended so you that i guess

[molten oracle]

One of the bigger problems with python imo is the strictness with consistency of indentation. The inability to mix spaces or tabs is not a big issue for the average programmer but for a beginner it's neither visible nor something easy to figure out.

[molten oracle]

Wsl2 is new so hyper v support exists mostly for backwards compatibility but WSL2 is so much better.

[topaz cedar]

Suggest me a way to develop a gui(any lang) in less time, that can run on 🍎, 🪟, 🐧. I hv to make a robot controller.

[simple canopy]

Sorry to everybody here, as I am going to suggest Electron… you want to run on iOS as well? If so I‘d suggest PWA builder… develop your UI using Vue.js, Svelte, React, Angular,…

And if it should run on mobile and desktop, this is as far as I am aware your only option… well Ionic would be as well, but that’s a similar idea

[molten flicker]

Just use Linux

[molten oracle]

IDK what your apology is for. Electron is far from the "best" objectively, but they specified they want to do it fast, and Electron is one of the easiest way to make good looking GUIs in a short period of time.

Tauri is up and coming as a better alternative but I'd argue it is still in early stages.

[simple canopy]

There are some people who hate Electron

[molten oracle]

Electron is bad, if its feasible to make native apps, always WAY better than electron.

But the issue of feasibility is why electron exists.

[molten oracle]

"Should I use x or y"
- "Use Φ"

[molten flicker]

React Native. It's supposed to have decent Gtk/libadwaita/fltk/qt/etc bindings.

[molten flicker]

Tauri is also a good option

[sudden cosmos]

electron is fine if one company uses it, once every company uses it like now your computer is unusable because you have 8 instances of chrome running. VSCode hardly even runs on my work VM they give me alongside firefox, if i launch a third browser like obsidian they all crash.

[sudden cosmos]

How is it controlled? If it's over the web the simplest thing is just an html page with embedded js. If you need actual native bindings then you might consider like Python+Qt

[topaz cedar]

It's using ROS with python, so it's native

[sudden cosmos]

oh the GUI will run on the robot itself?

[topaz cedar]

No, the gui runs on the computer over a TCP connection

[sudden cosmos]

i see yeah I think i'd lean PyQt especially since you're already doing python

[topaz cedar]

Maybe since the data is on the network, probably it can be controlled over a simple js, html based page i.e, electron or tauri.

Which one do u suggest?

Native or Web?

[sudden cosmos]

you can do TCP over the web but i wouldn't recommend it if this is going to production. you can actually just put whatever you want in an ajax request though. but using native sockets via electron or pyqt will be basically the same experience. electron is kinda complex though, with pyqt you just pick it up and start writing python none of that js/ts build system stuff

if you want like a WYSIWYG gui editing experience I've seen some things use Godot. haven't tried it for that but you get cross-platform exports and maybe the gui part will be less painful

[sudden cosmos]

but i'd suggest PyQt. if this is just for learning though you have a lot of options and can pick whatever is interesting: PyQt/pywebview, Godot, C#/.NET, Iced(Rust), Java, Electron/Tauri(JS), thug it out in C/C++, etc etc

[frozen flame]

Be careful with QT

[sudden cosmos]

why's that?

[frozen flame]

The license on qt makes it a real pain in the ass (or in the wallet, take your pick)

If you're using python and don't want to go open source, try kivy

Works with nuitka too if you want an executable binary

Also if you do want to use qt, use pyside since those are the official bindings (arguably better too)

[sudden cosmos]

ah good call, pyside seems to be LGPL which should be fine.

[nocturne galleon]

is there a vscode extension for visualizing large qml binding trees? similar to how you can view function callers

[nocturne galleon]

keep in mind that although the LGPL allows proprietary linking, it doesn't allow proprietary bundling unless you distribute the library's source code together with its binary, so you'd need to at least link (in the website sense) to qt's source code, and that has to happen for every qt version any version of your program ever ends up using

If you dynamically link against an LGPLed library already present on the user's computer, you need not convey the library's source. On the other hand, if you yourself convey the executable LGPLed library along with your application, whether linked with statically or dynamically, you must also convey the library's sources, in one of the ways for which the LGPL provides.
from https://www.gnu.org/licenses/gpl-faq.en.html#LGPLStaticVsDynamic

[keen sorrel]

I recommend first assessing the requirements for your UI before deciding on a framework. This can allow you to rule out UI frameworks that lack a particular widget/component/feature that may be desirable for your design.

[bitter obsidian]

Got a random make question. I'm having an issues building a project with a Makefile on a mac (vscode). I tried adding .RECIPEPREFIX = - so I could use a hyphen instead of tab for commands but make doesn't seem to like it and only let's me use tabs like so:

    @echo "Cloning repo.."```
Instead I would like to be able to do this:
```clone_repo:
- @echo "Cloning repo.."```
Am I doing something wrong?

[bitter obsidian]

Found my issue. I need to be on make 3.82 i'm on 3.81 https://stackoverflow.com/a/5898784/1973777

[next igloo]

Sometimes the CPU and GPU will render at roughly the same speed (especially if the GPU is using 64bit floats). In this case, what are the pros and cons of rendering on the CPU vs the GPU?

[peak acorn]

pov building a TOS infringing service

[nocturne galleon]

POV I don't care

[molten oracle]

Kinda should not be admitting it publicly

Gonna have to CC <@&750150305383186585> for this one

[grizzled steeple]

In case anyone needs some colorful icons to spruce up their visual resume, some Japanese dude is making some:
https://fxtwitter.com/sawaratsuki1004/status/1782079506083381657?t=jnmBhb0FAJz869PiLU_utA&s=19
(they want to migrate to GitHub, tho the repo is empty for now: https://github.com/SAWARATSUKI/ServiceLogos)

[hollow girder]

Hi everyone , iam a cyber security student and i need ideas for my final year project , it shouldnt be hard cus iam no very good , please suggest me some , Thanks

[nocturne galleon]

https://www.instagram.com/reel/C53wKC9ICt2/?igsh=MXF1MWdob2JrMm9lOQ==

growthhungry.academy

Your thoughts?🤣

#programmingmemes #careerinIT #careerinBigTech

Likes

566441

Instagram

😭

[keen sorrel]

Determine what to use by what resources are available. Your software could be running on some 64+ core server CPU with an RTX A1000 (equivalent to an RTX 3050 OEM) or a desktop with an 8 core CPU and an RTX 4090. It could be anything in between or maybe your software runs on one particular segment of devices that makes the answer easier for your case.

[next igloo]

My 11 year old laptop had an Intel HD 4000, so the CPU was roughly the same speed as the iGPU when rendering with 64bit floats; and I could definetly make the CPU pull ahead if I speed things up more with AVX instructions

[keen sorrel]

If you think that device is a good representation of the software's user base, then sure target that

[next igloo]

I also found it odd that you made the comparison between two Expensive hardware configurations

[keen sorrel]

It was completely arbitrary. Extremes to demonstrate a point

Unless you're running software for one very specific device, you need to implement a system at runtime to detect hardware capabilities and execute shaders on the appropriate devices for that hardware. Depending on what your userbase tends to deploy the software to or even just what hardware platforms you support, this runtime system can be more or less complex.

If it's software you're running personally on your device alone or primarily, then it's even easier. Use profiling to gauge what resources you have and tune accordingly.

[nimble sand]

Does anyone know JavaScript and could help me?

[mental panther]

https://dontasktoask.com

[nimble sand]

Ok thanks for nothing

[fresh valley]

Sure I know javascript reasonably well, but what exactly is your question? As Ethan was saying your message provides little of the required context

[nimble sand]

I wasn't trying to come across stand offish, I've been trying to get this running for about 6 hours and pretty tired. I have a HTML file which contains JavaScript, the "simple" premise is this:

It's an online chart that is akin to a stock market graph, but used by a barber shop to show their prices. Owner would like to publish it online and have customers check if the price has changed according to how busy he is.. hope I've explained that well enough!

I have the code to the point where it will do what I want it to do, it just has an annoying bug where I need to set the price twice otherwise it will always remember the previous price and jump back to it

It's not pretty, but it's somewhat functional

[fresh valley]

The reason they would need to set the price twice is that you are appending the previousPrice on line 116, but you set previousPrice on line 95 BEFORE you update to the user's new value

if that is indeed what you intend to do, in your update loop you will need to make the previous price be the current price after you've appended the previous price

@nimble sand

[nimble sand]

@fresh valley ok understood. I'm not sure why I even put previous price in there now. I think it was because I couldn't get the green line to continuously draw a line if it didn't know the previous price

I changed line 116 to currentPrice and it seems to be fine now

I know it isn't the cleanest code in the world, I have tired eyes and needed a fresh look so thank you for your help

[fresh valley]

glad that could help, it very much seems to me as one of those problems that you just overlook after writing too much code in the same session and seeing what you wanted to write rather than what you did. One thing I would mention is that when you start to include JS code into actual functioning websites instead of tech demos, you should use seperate files and load them in using the defer attribute. It makes the codebase far easier to understand for multiple people in development, and will generally allow for cleaner and more readable code.

[nimble sand]

I appreciate your insight. I appreciate that is probably some of the most basic code you've ever seen but I literally started reading about JS last Friday

[fresh valley]

Absolutely alright, everyone has to start somewhere and I have nothing better to be doing at 2am anyway

[elfin mica]

I'm trying to make a web app that generates images using Stability API but I'm getting an error:

the error is on line 10: const response = await fetch(import.meta.env.VITE_API + '/dream', {

Here's the front-end: ```javascript
import './style.css';

const form = document.querySelector('form');

form.addEventListener('submit', async (e) => {
e.preventDefault();
showSpinner();
const data = new FormData(form);

const response = await fetch(import.meta.env.VITE_API + '/dream', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({ prompt: data.get('prompt') }),
});

if (response.ok) {
  const base64Data = await response.text();
  const binaryData = atob(base64Data);
  const bytes = new Uint8Array(binaryData.length);

  for (let i = 0; i < binaryData.length; i++) {
    bytes[i] = binaryData.charCodeAt(i);
  }

  const blob = new Blob([bytes], { type: 'image/webp' });
  const blobUrl = URL.createObjectURL(blob);

  const result = document.querySelector('#result');
  result.innerHTML = `<img src="${blobUrl}" width="512" />`;
} else {
  const err = await response.text();
  alert(err);
  console.error(err);
}

hideSpinner();

});

function showSpinner() {
const button = document.querySelector('button');
button.disabled = true;
button.innerHTML = 'Dreaming... <span class="spinner">🧠</span>';
}

function hideSpinner() {
const button = document.querySelector('button');
button.disabled = false;
button.innerHTML = 'Dream';
}

and here is the back-end: ```javascript
import fs from "node:fs";
import axios from "axios";
import FormData from "form-data";

import dotenv from 'dotenv';
dotenv.config();

import express from 'express';
import cors from 'cors';

const app = express();
app.use(cors());
app.use(express.json());

app.post('/dream', async (req, res) => {
    console.log("--POST")
  try {
    const formData = {
      prompt: "Lighthouse on a cliff overlooking the ocean",
      output_format: "webp"
    };
    
    const response = await axios.postForm(
      `https://api.stability.ai/v2beta/stable-image/generate/core`,
      axios.toFormData(formData, new FormData()),
      {
        validateStatus: undefined,
        responseType: "arraybuffer",
        headers: { 
          Authorization: `Bearer `+ process.env.STABILITYAI, 
          Accept: "application/json" 
        },
      },
    );
    
    if(response.status === 200) {

        // console.log(response)q
    //   const data = response.json()
    console.log("---DATA")
    console.log(response.data)
      res.send(response.data)
      //fs.writeFileSync("./lighthouse.webp", Buffer.from(response.data));
    } else {
      throw new Error(`${response.status}: ${response.data.toString()}`);
    }

  } catch (error) {
    console.error(error)
    res.status(500).send(error?.response.data.error.message || 'Something went wrong');
  }
});

app.listen(8080, () => console.log('make art on http://localhost:8080/dream'));```
Any ideas?

[wooden shore]

What is VITE_API value

[elfin mica]

VITE_API="https://localhost:8080"```

[wooden shore]

Remote the s

From https

[elfin mica]

thanks. Now I got a new error haha

const binaryData = atob(base64Data);```

[wooden shore]

You probably have uri encoded base64

Hmm I had this issue recently. Print out base64data and see what it contains

[elfin mica]

like this?

[wooden shore]

That looks like it is json already