All boardsБлинчикиInternationalКрендельнаяЛес

#1 SciTech Thread

#976927

#1 SciTech Thread

1920×10803.67Mb00:08

Kpop science and technology here!

#976928
3000×20003.50Mb

mounting remote disks to use from application code is kinda super meh, so instead of that I thought of an http service to handle writes/deletes/links and a library to use it from code.

like S3 but local wdythink rubber duckie but also not complicated and slow like minioh


also I kind of want to move to non-raid to keep things max simple (and to unlock unlimited cloud backup) but how should I then handle expansion? my current thinking is adding a "bucket" argument to requests where each disk is a bucket so when one is starting to get full switch to a new bucket is it too simple even holy

#976933
1290×8602.78Mb00:06

>>976928

>

remote disks to use from application code

what application?

>

like S3 but local

local FS? or your own server?

>

also I kind of want to move to non-raid to keep things max simple

without RAID it will take more time to restore data if one disk suddenly goes down. with RAID you just change disk and everything continues to work

>

my current thinking is adding a "bucket" argument to requests where each disk is a bucket so when one is starting to get full switch to a new bucket

yeah sounds simple enough

>

unlock unlimited cloud backup

is it even a thing? I read a bit of https://forum.rclone.org/t/unlimited-alternatives-to-google-drive-what-are-the-options/36529 recently and there're almost no good options

#976939
2207×3311839.73Kb
1990×2983529.05Kb

>>976933

oh hey rubber duckie

>

what application?

youtube and image archivers. kpop!

>

local FS? or your own server?

yeah the computer with the disks would also host the http service that would then write the files to disk. a poor mane's limited bucket size objekt storage

>

without RAID it will take more time to restore data if one disk suddenly goes down

it's just kpop so doesn't really matterino

>

is it even a thing?

looks like amazon drive is gone but backblazeh is still there

a simple 10 dollarinos a month to backup a single computer and it's drives

#976941
4096×40963.18Mb

>>976934

duckie yeah i tried it for a while but it was very unlit

small notes for the duck from last night


smb/nfs mounted disks:

>

unnecessary file permission memes

>

question mark whether its actually connected or if i'm writing to the unmounted mount point dir

>

if application is dockerized prepare another layer of mounting and file permission memes

>

often an expensive, high intelligence raid setup like a duna

minioh:

>

performance slow

>

files not stored as real files but split into weird pieces

>

no softlinks

a small http service that writes things onto disks so you only need to send a post request to an ip address:

>

comfy af

>

small and cute like a yuna

#976942
1125×2000402.22Kb

>>976939

>

yeah the computer with the disks would also host the http service that would then write the files to disk. a poor mane's limited bucket size objekt storage

can't you use something even simpler like SFTP then?


>

youtube and image archivers. kpop!

unraid might come handy because of simple server management & raid 4


>

but backblazeh is still there

backblaze backup is only for windows/macos and with a lot of limitations like not syncing external drives. no free lunch

backblaze b2 is pretty expensive per TB

I would recommend OneDrive: 6TB per family account and you can buy keys for ~$56/year, so it's about $0.8/TB/month. that's what I personally use for off-site backups


>>976941

I can't see how HTTP service with API is simpler than SFTP


>

mini ooohhhh

#976945

also I don't understand why do you need some HTTP server at all

if you have computer with connected disks, then just ran archiver program from that computer and everything is just a local file

#976946
1350×10802.59Mb00:08

>>976928

>

but also not complicated and slow like minioh

what about MongoDB GridFS

found discussion about some guy using it for >150TB >1B of objects on XFS


overkill for kpop tho, you probably won't find 1B of pics of kpops in the whole net

#976949
2731×40961022.05Kb
2731×4096961.16Kb

>>976942

>>976945

yeah computer needs to be win/mac with local disks to enable unlimited backup, but I also dont want to code against those platforms so

>

6TB per family account

my yt collection is already 8tb although I'm looking into slimming that down with better review tools so I could move fully onto SSD

>

SFTP

cool

although I don't want to cd mkdir rmdir etc on client side so I would have to write a wrapper. but then I might as well write the 200 lines of handling server side to make client library as simple as possible

#976950
1364×2048247.45Kb

>>976946

>

GridFS divides a file into chunks of 255 kB with the exception of the last chunk

looks like another minio, too fancy for me dumb brain

#976951
1000×1500173.13Kb

how is rubber duckie so smart

#976952
1920×10802.36Mb00:03

>>976949

>

my yt collection is already 8tb

you can make several accounts

tho I'm not sure you need off-site for yt archive, seems like not so important. might just use separate computer for backup in a different room for safety

>

on client side so I would have to write a wrapper

>

server side

do you have multiple computers? why do you split archiving into client/server?


>>976950

seems like you will need those if you have millions of small files

if your files are big then normal FS is better


>>976951

I'm not a duckie

#976955

https://kkovacs.eu/cassandra-vs-mongodb-vs-couchdb-vs-redis/

>

Best used: If you want something Dynamo-like data storage, but no way you're gonna deal with the bloat and complexity. If you need very good single-site scalability, availability and fault-tolerance, but you're ready to pay for multi-site replication.


oh Riak seems interesting for distributed blob storage of files

#976974
1772×26571.24Mb

>>976952

>

why do you split archiving into client/server? do you have multiple computers?

oh I just mean the server part of the local objekt storage versus client library I would use in my archiver code.

but yeah I currently have the computer I use for computer stuff, the home server that has a bunch of VMs on it, and a nas with spinning disks . but I'm thinking of replacing the nas with a regular ol' computer with SSDs


the nas has a one disk redundancy so I guess that's something, but it would be too expensive to backup offsite

>>976958

I tried minio for some time but hated it it's so slow and because of the chunking backing it up would mean running some slow bottom command to get the files back into real files

surely regular old SSDs can handle even millions of images if I don't put them all into a single directory

>

tho I'm not sure you need off-site for yt archive, seems like not so important

it's something I've also thought about. going no-redundancy for youtube and only really backing up the videos that return 404 since everything else can be redownloaded

#976987
738×1080921.76Kb00:01

>>976974

>

and a nas with spinning disks . but I'm thinking of replacing the nas with a regular ol' computer with SSDs

can't you run archiver (I guess yt-dlp) on NAS? and all your files will be available locally, no need for client/server


>

I tried minio for some time but hated it it's so slow

I read not very good reviews about it, seems a bit toyish. other solutions like Ceph should be more mature but need to google more, maybe there's even something better

tho in your case all of that is definitely overkill. I was just having some fun looking into that stuff


>

surely regular old SSDs can handle even millions of images if I don't put them all into a single directory

it's not about physical disk, it's about particular FS implementation you use

millions seem ok (for modern FS) if splitted into ~1-5k per directory yeah (but there's no definitive answer on the web about performance penalties)

but billions could be too much

also with FS you need to implement sharding/replication manually


>

going no-redundancy for youtube

that's what I'm doing

but I would rather do on-site backup of all data to separate computer and off-site only important if I had spare computer and more disks

#977009
1365×2048496.67Kb

>>976987

>

can't you run archiver (I guess yt-dlp) on NAS?

initially I did, don't remember i moved it off i guess it was too slow (very low powered synologhy) and also i needed a more complete vpn/proxy solution

>

millions seem ok (for modern FS) if splitted into ~1-5k per directory yeah

>>977008

average filesize of images I've downloaded is 720K

#977012
380×6401.18Mb00:18

>>977009

>

i moved it off i guess it was too slow (very low powered synologhy)

but yt-dlp shouldn't be CPU-intensive, it basically just dumps data from TCP stream to disk even tho it's HTTPS still should be very fast, faster than disk speed

or maybe connect disks from synology to server with VMs?


>

average filesize of images I've downloaded is 720K

how many photos do you have?

#977044
1000×10801.95Mb00:06[THE 미니웨이] tripleS (트리플에스) [THE SHOW 230228] [ZnD7xRuFdI8]

beauty type kpops have a more oval head while cute type are more round

science? or just my head canon

#977056

>>976974

>

I tried minio for some time but hated it it's so slow and because of the chunking backing it up would mean running some slow bottom command to get the files back into real files


minio is basically object storage abstraction over FS

>

On the other hand, MinIO seems to store each object in a directory and two files. So the overhead is even higher.


the only use case is replication and data redundancy but with low performance for basic tasks it seems uselless


if you have billions of small files and file system isn't an option then something like Ceph seems a better option

or if you don't want to replicate files by yourself, Ceph has automatic data redundancy too


tho in case of single server it seems easier just to backup all files instead of those complicated erasure coding schemes like in Minio/Ceph

#977066
1365×2048246.15Kb

>>977012

true maybe I messed around with something else

#977070

>>976974

>

backing it up would mean running some slow bottom command to get the files back into real files

not necessary btw, you can just backup the minio's data directory, all your data is stored there

#977071

>>977012

also around 100k fotos but it's a mess currently hehe need to figure out a better search setup using tags and maybe face recognition if i'm able

#977076
1064×1596165.21Kb

>>977072

simplest option of all the local S3 thingies I could run in syno

#977086
748×748404.21Kb

>>977078

true I'm currently writing a babby's first objekt storage server but with real files instead of chunks

fn upload(UploadArgs { key, data }: UploadArgs) -> Result<()> {
    std::fs::create_dir_all(key.parent().unwrap())?;
    let mut temp = NamedTempFile::new()?;
    temp.write_all(data)?;
    std::fs::rename(temp, key)?;
    Ok(())
}
#977293

moved docker.img from vDisk to folder because I think it was wearing my SSD

fucking temporary useless files that are written to disk for some stupid reason

#977412

What programming language should I learn in 2024 for employment?

#977446
3000×2000573.62Kb
3000×2000447.93Kb
3000×2000796.88Kb

>>977364

is this something i need to worry about if not using unraid

i'm 4 btw

>>977412

prompting LLMs like chatgpt to program for you? other than that idk i'll let smart-nim to answer

#977453

>>977446

>

i need to worry about if not using unraid

I think no, you probably have /tmp as tmpfs already. in unraid docker setup is a bit weird, via additional btrfs FS mounted as loop .img and a lot of containers write logs to SSD


might run

#!/bin/bash
date=$(date +'%Y-%m-%d %H:%M')
wear=$(smartctl -A /dev/device | awk '/Total_LBAs_Written/{printf "%.3f %d", $10/2/1024/1024, $10}')
echo "${date} ${wear}"

like every day for checking total TBW (GBW in this case)


I'm running it now with inotifywait to check writes and nothing writes anymore except cron tasks I set


>

i'm 4 btw

4 years old?

what OS is installed on your VM server though?


>>977412

dunno, you will probably need to learn many anyway, it's not hard when you know few

might start with Python

#977456
1080×1920408.29Kb

>>977453

windows for that hyper-v comfy

although virt-manager also looks good so might switch to debian at some point

#977551
1800×550212.32Kb

fancy

i'm too good at breaking stuff to dare go for it but it looks like a way better synologyos

#977555

>>977551

yeah, it's pretty nice, although have its own issues

just moved my cache pool to zfs, all migration workflow worked very nice


but now I'm thinking about XFS vs ZFS on main array

and ZFS + BTRFS docker img vs ZFS + docker directory vs BTRFS + docker directory

on cache pool


but I think I don't like BTRFS at all but ZFS + docker directory also have its own problem

too many options

#977842
1350×2400586.62Kb
1350×2400327.00Kb

#977879
1060×1350660.10Kb

>>977876

1 billion imageru served from home server huh

cool project

#977880

>>977879

and they didn't recommend him anything better than FS

this is sort of highload case, I guess you won't find much in simple discussions and hyped open source, need to write by yourself or experiment a lot

#977881
>

I know though that making (uncompressed) tarballs of small files, then storing indexes into these tarballs on fast storage (maybe your database) would help though, I did that in one project. Alternatively, just get rid of the file system (crazy, I know), and write your tarballs directly on block storage. This is a viable choice only if your dataset doesn't change, obviously. I wasn't that crazy though.


oh, I know a highload project which does that

seems like a good idea to minimize IOPS requirement

#977927
2560×1440861.20Kb

>>977908

billions of files huh

maybe open up some 60fps fancams, hit alt+s and let it burn

#977947
2000×30002.48Mb
620×61892.21Kb

babbys first face detection

but hwere i now put this bad boy supposedly i need to create a mystery embeddings vector that contain the absolute essence of yeoni

#978000

>>977947

>

embeddings vector

yeah, but it's not very precise

better to download 1B pics of kpops and train by yourself

#978015

>>978002

dunno

Ubuntu is a bit stupid with rolling because you always have outdated software or need to use snaps. but it means you have some amount of outdated base packages and some are new and there's 2 ways to install something. not very elegant

in Arch something will always break

Fedora dunno, seems too experimental

Ubuntu-based doesn't seem better than Ubuntu itself, same as with Debian


Manjaro maybe?

or maybe there's some new hyped distro, haven't looked into it for a while because I don't have that problem anymore

#978020
4500×2999694.71Kb

>>978002

how about Windows + WSL?

#978039
394×61151.05Kb

>>978035

>

Creating embeddings for 1 BILLION images

the absolute madman

#978059

>>978054

dunno, maybe naver has some

it should be trained on asians I believe

#978090

>>978035

>

I have a billion images I need to create embeddings for which are stored on 100tb of flash I have in my homelab. I am using mongo to hold a reference to each locally stored image

so how do you think he did it in the end, sounds like a path to a regular ol file to me

#978095

>>978090

yeah, seems like it

but storing and serving are different tasks, it's ok if it's a bit slow when you just train network on those files, training will be slower anyway

just need to keep an eye on inode count

#978199
662×8288.76Mb00:32

#978491

>>978002

probably doesn't matter much

I don't really like 'stable'/non-rolling releases, it's a pain to wait for fixes you need

you can check if you can install software and drivers you're relying on sanely

I've been using the same install of Arch for four years and stopped getting breakages after ~1 year, when drivers caught up to my hardware

#978550
550×1259.13Kb

background: i'm basically a retard

realisation: create ml by apple incorporated works?

#978551
1162×2048216.74Kb

should i try gathering a bunch of yeoni pics and a bunch of not yeoni pics and put them in create ml

yeoni-or-not-yeoni classifier

#978820

>>978491

I think Fedora > Ubuntu > Arch > Suse

don't like when too many updates so Arch is on 3rd

but Ubuntu's outdated base system might be inconvenient, Fedora looks better

but I haven't used Fedora much, not sure

also heard good things about Suse such as Tumbleweed, but it's much less popular so less howtos

#978903

>>978820

yeah the first three are safe recommendations imo

#978947

I think a mix of different models is optimal


something like:

beta/unstable/git-master for things that give fast and important updates such as compilers, encoders

rolling for most other CLI/GUI tools such as browsers, editors, docker

stable for core system libraries which shouldn't break your workflow just when you clicked update


you would never want to install browser or yt-dlp from Ubuntu's package manager because you want them to be always updated to the latest

nowadays most GUI software in Windows/macOS can update itself to latest versions so it's not a big problem, but a bit harder with CLI tools which are just on github as source code

#979313
>

With a Zimaboard plus PCI SATA plus low end 16T HDDs we can get 1PB for 64 disks * $250 + 13 boards * $240, or for $19120

#987324

https://www.theverge.com/2023/10/26/23934216/x-twitter-bank-elon-musk-2024

>

It’ll be on our platform. Money or securities or whatever. So, it’s not just like send $20 to my friend. I’m talking about, like, you won’t need a bank account.”

people have already gotten fuckd by apple card, imagine a tw*tter as your bank

this whole saga has to be some ultra elaborate gigatroll

#988184
1080×1440219.46Kb

i tried chromeos distro. native apps (non web app) might work fine but where is drag and drop?? i need it for mpv

anyone know how to get drag and drop?

#988614

Is there like a clamp or mic stand i could buy to also occationally use to keep a sauce bottle (or whatever) upside on top of a plate for a long time to empty it

#988645

>>978947

>

CLI tools

you can use choco, winget (windows)

or macports, brew (macos)

#988733
1000×6664.36Mb03:08Bae.Suzy.full.171056

I can't read postgres explain execution plans

it's over

#988760
1280×7202.53Mb00:03
>

Price for 8TB of TLC generic is currently 170 USD. QLC is cheaper than that

where samsung 8tb qlc is motherhecking 380 USD

#989162
2048×2048360.10Kb

>>989110

chatted with my chingu mr. copilot and he helped me increase work_mem at least the query works now though still bit slow at 2 seconds

#989271

>>989162

>

copilot

do you pay for the subscription? I thought about using it but don't want to pay for the basic stuff


also, help me choose icons collection: https://icones.js.org/

there're too many options

#989278
>

Congratulations! You are eligible to use GitHub Copilot for free.

wait

#989351
1536×2048428.25Kb

>>989271

yea some code kinda repetitive to write without mr copilot

the chat part can be dumb sometimes since it cant read the file as context if the file is too long similar to chatgpt


idk abot icons heroicons seems simple

#989378

>>989351

do you have enableAutoCompletions set to true in vscode?

I found you can trigger it with Alt+\ manually, not sure if I want to see those auto completions all the time, it distracts me


maybe need to get used to it

also afraid about sending all my code to server all the time. not like I'm not googling every line of code anyway

#989409

--proxy-pac-url not working in VS Code

so I have to proxy all traffic of VS Code via http_proxy


or maybe route github.com via VPN system-wide but I don't like it much

#989419
730×830973.87Kb00:02

>>989409

eligible for free but not able to use due to location? dang

>>989378

looks like yes (i don't change defaults much)

it takes a bit of times before suggestions plus can always press esc so its fine


but yeah it sends everything maybe one day localpilot or something will have nice experice and i can use it instead

#989420
% du -sh node_modules/@tabler
166M	node_modules/@tabler

#989424

>>989419

>

eligible for free but not able to use due to location

tough times

well, should work fine via proxy too


I'm using wireproxy now, comfy tool (localhost[HTTP PROXY] → Wireguard → remote host)

#989459

>>989440

oh no, in ionic it's jus 3 version of the same icon so just 400 icons each


will try tabler then. it's all so tiresome

there's also material icons by google

#989479

>>989378

trying it, some suggestions seem to be nice


but can't do even simple stuff like "fix framework imports" when I ask it

maybe I'm doing it wrong

#989905
1440×1920401.78Kb

oops wrong thread

>>989479

i havent used it much but its probably just a regular chatgpt 3 with currently open file as context

>>989808

gotta just wait until cool people solve it and then make a one click vscode thingie i dont have computer resources for LLM anyway

#989908
2124×2124696.98Kb

>>989907

it's the ml hyeseong what can you do

#989909

why stable diffusion is so shitty at generating logos? it took me just few tries to generate nice logo in DALLE3, but I spent all night yesterday with stable diffusion and that shit only generated complete garbage


people can't stop shilling their shit huh

#989949
720×12803.05Mb00:06

I like Copilot nim now, it really makes me write code faster

not drastically, but maybe by 20% at least


just need to stop getting distracted by suggestions too early. i.e. I have to write something meaningful to make it autocomplete it. but it tries to autocomplete even when I wrote just single word and my stupid brain looks into that autocompletion and loses time and focus

#989967
2483×2160383.07Kb

>>989949

im not sure if i notice the early suggestions anymore, maybe sometimes so shud be fine

#989974

>>989967

>

if i notice the early suggestions anymore

it's like when you type with someone else watching at your screen and commenting it out loud

#989975
600×600508.71Kb00:02

>>989967

new M1 MacBook Pro

rich girl

#990122

>>989479

>

fix framework imports

oh, I'm stupid, you can just press Ctrl+. to make vscode automatically fix imports

#990411
738×1080921.76Kb00:01

used copilot for few days

I'm a bit scared now, sometimes it's too good at predicting my mind

#990702

>>989909

can you see the prompt chatgpt creates for you for dalle3?

what if you put that into stable diffusion

#990703

>>990702

>

the prompt chatgpt creates for you for dalle3

dunno how to see it

also, do they use chatgpt directly? has to be more complicated machinery

#990704

>>990703

apparently they take your prompt and turn it into some detailed prompt that works better maybe that's most of the magic between dalle2/sd and dalle3

>

When prompted with an idea, ChatGPT will automatically generate tailored, detailed prompts for DALL·E 3 that bring your idea to life

#990706

>>990704

I tried to ask ChatGPT for prompts for SD but it didn't help

but prompts from the google for logos also gave me shitty results


maybe there's something wrong with the model I've used

#990723

I read about Tailwind and Figma and now not sure if I need it

learning new tools is so complicated

#990814

I hated tailwind at first glance but now I'm starting to like it

#990865

>>990862

for me the main pros is because I can just simply copy paste styled HTML components from the internet

I'm suck at design


otherwise BEM seems like a much more solid solution, if not prototyping of course

#990866

HTML + CSS is solved with tailwind

HTML + JS is solved with JSX frameworks


how to solve JSX components issue

there're a lot of libraries of components like material-react but not sure, seems like you often want to customize? or use with non-react?

#990870
1466×2048365.39Kb

https://www.material-tailwind.com/figma

interesting


you can create WYSIWYG design previews in Figma using some UI kits

then quickly add those components from JSX library to your web app code


seems to be better than trial and error way of designing web app directly, without design app

#990874
1440×1800269.08Kb

>>990870

but then you would design without content on figma

isn't that way more difficult


isn't it better to put all the lay out all the content without styling and then at some point style them once you know what kinda stuff goes where (idk i cant desing)

#990875
720×12801.72Mb00:15

>>990874

>

isn't that way more difficult

dunno, depends on a website I guess


sometimes I have rough ideas of the web UI I want to implement, but it's kinda long process to do all the HTML/CSS/JS code

so when I implement stuff out of my head in web app, it might be shitty UI and UX and I will have to re-do a lot again


with Figma you can iterate on prototypes faster and get idea about final UI and partially UX, so you can just implement the final version


maybe I'm just too slow at prototyping in JSX code tho?

#990891
913×790529.10Kb

everybody wants your money nowadays


figma: subscription for pro use

sketch: paid

tailwind components: paid

material components: paid

figma files: limited preview but everything else is paid

#990894
1536×2048289.60Kb

>>990875

oh you know figma i guess that works too then


im to disabled myself and can only work with what mr. browser gives me

#990928
1406×18040.19Kb

#990946

>>990891

i think it's cool people are able to create a job for themselves like that wish i coud do it

#990948
1280×72079.77Kb

>>990946

I think the root cause is greedy monopolist corporations like Apple, Intel and Microsoft


they sell their software/devices with a huge margin because greed

but tons of people use their products (because monopoly) so to make for living they have to milk customers as hard as they can too


fucking stupid capitalism with a rat race

#990978
1080×1080383.91Kb

>>990948

i guess because software interfaces are not exchangeable they can charge whatevermoney makes switching too difficult

idk how that relates to someone making a some ui pack tho, at least most of those appear to be one time purchases, in case you need one


if i were using react i would probably try mantine or material-ui and build my ui around it instead of trying to design my own they are free too i think

#990987
930×1200135.93Kb

>>990978

>

if i were using react i would probably try mantine or material-ui and build my ui around

I've started with a bit cooler framework, but now I think maybe it was wrong because a lot of libraries are made for react


but with those Tailwind templates it shouldn't be a problem because JSX is the same


though in some libs like https://www.material-tailwind.com/docs/react/button they provide convenient wrapper components

for those horrible Tailwind classes mess https://www.material-tailwind.com/docs/html/button#button


so I'm still thinking how to deal with it

maybe I should port those components https://github.com/creativetimofficial/material-tailwind/blob/main/packages/material-tailwind-react/src/components/Button/index.tsx

to my framework, they look simple. but that would be tedious


maybe should have just started with React. but it's not cool


>>990979

>>990983

it's interesting that it also uses Tailwind CSS framework, so I can copypaste that to my project too. need to try, looks pretty cool

there's waitlist tho


so tailwind looks like a bit improved inline styles. the benefit is that your HTML and CSS are connected together so you can't break them by renaming some class or changing markup a bit. also can just copypaste single block of test, instead of HTML + CSS separately. so in old days we had HTML, CSS, JS all separated, but nowadays it's JSX + Tailwind all in one file

but it still looks messy, like old WYSIWYG HTML generators


also I'm a bit scared about those AI solutions, it may soon replace devs

#990991
1490×10026.22Mb00:13

>>990987

>

for those horrible Tailwind classes mess

on the other hand you own the full code so can do changes very fast

without looking what are the defaults of that framework and how to change them the way you want

but it's if you want to change and know what's the final look should be. if you're ok with default themes then React component solution is cleaner, and also faster to prototype. uh oh


>

there's waitlist tho

maybe Copilot/ChatGPT can generate them too?

now I'm starting to get the full idea of Tailwind too. it's easier for AI to operate in, because everything is combined in a single chunk of code, you don't need to context switch between template and styles and keep track of connections. your context can be very small when you generate this


very cool if you think about that. did creator of Tailwind have AI in mind?

#990993
1440×1800289.43Kb

Thank you for joining the waitlist for v0 by Vercel Labs.

Please stay tuned for an invitation to get started.


#990995
1238×117071.08Kb

>>990979

it's interesting that you chat with it and ask to change bit by bit initial generated code


it probably uses ChatGPT API underneath tho? just a tiny wrapper on top of OpenAI's API

#990997
<button class="inline-flex items-center justify-center text-sm font-medium ring-offset-background transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2 disabled:pointer-events-none disabled:opacity-50 h-10 w-10 text-zinc-100 rounded-full hover:bg-gray-600 hover:text-zinc-100">

can browsers render that mess efficiently tho?

probably yes, but still looks horrible

#991035
326×91875.78Kb

tailwind brings semantic HTML tags back

classes are basically unreadable so you have to use nice tags instead to navigate in that mess

#991042
2182×1218480.60Kb

I've seen enough

#991046

>>990995

>

remove bg-opacity-50

might as well use keyboard at that point dumbass example

>>991042

in this case you'd make components out of label and input the styling gotta live somewhere

#991157

>>991046

>

the styling gotta live somewhere

yeah but those classes are horrible to navigate it, with normal CSS props it would have been easier

#991250

>>991237

>

<details>

interesting


>

focus on content

I'm just thinking it's much faster to iterate on UI with a library of components when you don't need to reinvent everything from scratch

probably should have just picked React


of course it's not very cool to use existing solution for design but most people don't really care how efficient or unique the site is

but if you make very simple half-assed implementations for dropdown/popover/etc with basic markup/styles people would probably notice it and will think it's ugly


although it's not really hard to make JS logic for dropdown, it's mostly the styles and UX which should look nice and it's too long to do from scratch

can just copy from Tailwind templates but why not copy the whole React component in that case

#991271

>>991250

>

can just copy from Tailwind templates

oh this looks nice https://flowbite.com/docs/components/alerts/


>

why not copy the whole React component in that case

harder to modify if you don't like something I guess

just need to be able to create those components fast enough

#991405

>>990978

>

mantine

oh it also looks nice but depend on react and components aren't look easy to port

also should be harder to modify the style to your taste because you install components as a library and had to override the classes on top which isn't very convenient (not like most people would want to customize it much tho)


flowbite (or maybe another similar Tailwind collection, haven't checked them much) looks best so far - you own the full HTML with Tailwind classnames which you can change however you want

and they also have small simple vanilla JS implementation for components like dropdown/accordion which should be easy to port to any framework

https://github.com/themesberg/flowbite/blob/main/src/components/accordion/index.ts

#991408
1200×1920339.20Kb

>>991405

>

you own the full HTML with Tailwind classnames

seems like it's the best solution for custom sites nowadays


it gives you HTML template freedom of Bootstrap (with React components you usually can't change shit in template, it's all abstracted out in separate library in node_modules which is bad for UI stuff)

and also it gives you easy way to modify look of those templates (in Bootstrap you just have high-level classes like dropdown)


but not as easy as going with React and installing some component library and just importing the pre-made stuff into your project of course

#992254
1080×108058.76Kb

why does mpv crash me computer once every 3 months black screen and static sound until i turn off teh computer

#992268

>>992240

>

The not-obvious path forward -- which our elderly users cannot find -- is to tap "December 2022", which pops open this rolodex-type thing:

dunno, it's probably me, but I instantly noticed it as a way to enter the date quicker (and no, not because I remember it)


just the pure logic: designers must have the case in mind when you need to go back by 20-30 years or more, so there must be button somewhere. and that caret is the only one visible


but elders probably can't think that way

#992269
2572×74284.49Kb

>>992268

on the other hand, what would you use for the save icon nowadays?


zoomers don't know what floppy is, but other attempts to portray "save" are even worse

you can add "title" to the tag ("click to save") but it kinda sucks too

for me floppy is the best, but it's just me

#993193
1308×21609.37Mb00:10

am i the only one who has problem viewing 1080p youtube videos with mpv

if i try to mpv directly from youtubes usually i only get audio no video. if i download the video first it works, and 4k videos also always work

#993260

>>993193

>

if i try to mpv directly from youtubes usually i only get audio no video

might be because video chunks failed to download, check console for errors

I often have that with youtube videos too, especially when I do seeking to not yet buffered position

#993267

>>993260

>

check console for errors

oh right

[ffmpeg] Not detecting m3u8/hls with non standard extension

maybe its a hls thing and I should just change config to always prefer vp9

#993277
778×1023119.66Kb

>>993269

why not youtube premium HD (AVC)

VP9 compression artifacts on 1080p are horrible tbh, it becomes bearable only at 2160p because bitrate is much higher. maybe on 1440p and 1080p60 too, haven't checked for a while


I can't understand how people not notice that VP9@1080p is garbage and google just let it be that way

#993330
1080×19207.22Mb00:14

>>993277

same problem when trying with mpv directly it seems i just want a comfy setup

downloads at 1080p as mp4 for sure

#993342
>

i dont wanna wash me teeth its so boring

we have the solution for you: blizrush 7.0 - the revolutionary toothcleaning-sponge

>

#993359
1536×2048437.44Kb

>>989162

actually heck trying to optimize this thing

a materialized view will surely set me free

#993550

>>993359

went from 2-15 second queries to 100ms queries and 2 minute refresh

#993566
1365×2048514.81Kb

>>993553

working on my kpop archiver

my dumbass normalized schema is a bit slow to query so gotta resort to other options

#993567

>>993566

>

is a bit slow

how many data do you have

why are you talking about minute delays, it should be instant, you aren't google datacenter

#993634

https://nitter.net/multikev/status/1724908185361011108

https://github.com/tldraw/draw-a-ui

>

You are an expert web developer who specializes in tailwind css. A user will provide you with a low-fidelity wireframe of an application. You will return a single html file that uses HTML, tailwind css, and JavaScript to create a high fidelity website. Include any extra CSS and JavaScript in the html file. If you have any images, load them from Unsplash or use solid colored retangles. The user will provide you with notes in blue or red text, arrows, or drawings. The user may also include images of other websites as style references. Transfer the styles as best as you can, matching fonts / colors / layouts. They may also provide you with the html of a previous design that they want you to iterate from. Carry out any changes they request from you. In the wireframe, the previous design's html will appear as a white rectangle. Use creative license to make the application more fleshed out. Use JavaScript modules and unkpkg to import any necessary dependencies

bit different since they use api directly instead of chatgpt but i wonder if i wrote like this would i get better answers than just asking with a single sentence

#993637
632×10612.45Kb

risky but cool

#993641

>>993634

btw, I found one issue with tailwind:


with CSS any changes of styles keep your app's state completely as it is, so you can prototype faster

but with tailwind, all your styles are inside JSX components, so if you want just to add 1px fix to the border or something, it will trigger app update

there's HMR, but it doesn't work good enough, state changes when you re-save your component's source file also I often have to Ctrl+R because HMR fucks up with the state and there're many errors in console (even though the app is completely fine when you refresh)

#994216

>>993641

sounds true

components are probably best developed in storybook so you have hardcoded state for various situations

#994217

>>994216

>

storybook

oh, I looked at it, but seems overkill for simple apps where you don't have many components


I mean you will rather modify components for your specific view, instead of overgeneralizing them

#994218
1440×1440229.96Kb

>>994217

i havent used react much yet but i did for a small browser extension and it was comfy its just a story file with some broilerplate and your state situations

would have been extremely uncomfy to test all the archiving states without it

>

author info

>

subscribed/not subscribed/ignored

>

subscribing yes/no

>

post info

>

archived/unarchived

>

archiving yes/no

>

server response

etc

#994220

>>994218

>

would have been extremely uncomfy to test all the archiving states without it

hm, maybe

but I always retest app as a whole to be sure that the overall UX is perfect

maybe I'm spending too much time on it

#994227

what do you think about Open AI firing their CEO?

#994242
1440×1440195.81Kb

>>994227

top of the news, curious to see if it actually matters one way or another

#994243

>>994242

google also postponed gemini

openai seems to be focused on ai safety now


no AGI soon

#994515

>>994304

wtf, I recompiled the program myself and now it 20 times faster

#994539

oh no, the name on npm is stolen

#994544

>>994542

meh, it makes npm install and package.json ugly


ended up with smaller name than I planned. maybe it's for the better, looks cool

#994545

>>994304

>>994515

checked on iphone, my implementation is 2-3 times slower than on notebook there

but now it's fast enough


but I've also removed one feature to shrink the size, might need to add it

#994546
748×748404.21Kb

#994575

I don't understand the eyes emoji people use extensively to react on github issues and comments

wtf is that even supposed to mean

#994577
603×611113.14Kb

>>994575

>

mean various things like “you are seen,” “I am watching,” or “Intriguing!”

#994606

webassembly is faster than the native tool in my console

how is that even possible


maybe shell is too slow

#994610

also I put it in web worker because it was 150+ms

but now it's 10ms, should you even bother with such a short UI thread block time

#994682
>

feel the AGI

#994710

lateral joins are so comfy for aggregating related stuff compared to the aggregate and group by every other column experience

but so slo

#994850

people talk about AI and AGI, but if you look at the state of projects at github, it's all developing fucking slow with tons of issues created years ago, nobody had time to fix them


I can't see computers replacing humans soon

or maybe it will be completely new code, not the already created mess that's managed by humans by hand, at the end of 2023

#994852

https://web.dev/articles/tbt

>

The main thread is considered "blocked" any time there's a Long Task—a task that runs on the main thread for more than 50 milliseconds (ms)

>

If the task is long enough (anything above 50 ms), it's likely that the user will notice the delay and perceive the page as sluggish or janky


barely found some heuristics

in my case it should be less than 50ms, so async worker isn't necessary

but dunno, what if slow old android?

#994857
>

MineDojo: open framework that turns Minecraft into an AGI research playground. We collected 100,000s of YouTube videos, Wiki pages, and Reddit posts for training Minecraft agents.

#995071

fucking retarded HTTP2 which fucks all links from the same domain if there's packet drop

retarded idiots who shilled that google shit trying to convince everyone this is somehow superior protocols


why it's all so retarded

#995074

>>995071

hm, it was HTTP3 which is UDP and should be fine

maybe Chrome is shit

#995075
1280×7202.08Mb00:07

>>995071

its also extremely hecking difficult to handle in wireshark talk about troubling times trying to connect request to a response


though with http3 you cant see or do anything at all in sharkie so gotta disable it in browser feature flags

#995079
2048×1364172.77Kb
>

sharkie mentioned

#995118
1048×9901.34Mb00:02
>

HTTP/2 is enabled by default and domains on Free plans cannot disable it.

#995130
640×600357.73Kb00:02

>>994610

>>994852

>

32.100ms

oh yes, I've found a way to make it 30+ ms with complicated query

so on a mobile it would perfectly be 50+ ms

so Worker wasn't useless

#995307

how do people use discord, it's a fucking mess

my brain hurts after just 5 minutes of using it, fucking information and visual noise everywhere

#995312

my google searches "how to disable X in discord" show that a lot of people having struggle with the noise

but devs probably keep all those annoying features so that normies don't miss something

#995519

>>994710

actually left join aggregate and group by every other column is still ass

gotta make the actual query a subquery, then left join and aggregate outside of that. took me from 13 seconds to 200ms


i'm le machecking stupid

#995521
1280×72079.77Kb

the absolute state of npm 

  ├─┬ d3@7.8.5
  │ ├─┬ d3-array@3.2.4
  │ │ └── internmap@2.0.3
  │ ├── d3-axis@3.0.0
  │ ├─┬ d3-brush@3.0.0
  │ │ ├── d3-dispatch@3.0.1 deduped
  │ │ ├── d3-drag@3.0.0 deduped
  │ │ ├── d3-interpolate@3.0.1 deduped
  │ │ ├── d3-selection@3.0.0 deduped
  │ │ └── d3-transition@3.0.1 deduped
  │ ├─┬ d3-chord@3.0.1
  │ │ └── d3-path@3.1.0 deduped
  │ ├── d3-color@3.1.0
  │ ├─┬ d3-contour@4.0.2
  │ │ └── d3-array@3.2.4 deduped
  │ ├─┬ d3-delaunay@6.0.4
  │ │ └─┬ delaunator@5.0.0
  │ │   └── robust-predicates@3.0.2
  │ ├── d3-dispatch@3.0.1
  │ ├─┬ d3-drag@3.0.0
  │ │ ├── d3-dispatch@3.0.1 deduped
  │ │ └── d3-selection@3.0.0 deduped
  │ ├─┬ d3-dsv@3.0.1
  │ │ ├── commander@7.2.0
  │ │ ├─┬ iconv-lite@0.6.3
  │ │ │ └── safer-buffer@2.1.2
  │ │ └── rw@1.3.3
  │ ├── d3-ease@3.0.1
  │ ├─┬ d3-fetch@3.0.1
  │ │ └── d3-dsv@3.0.1 deduped
  │ ├─┬ d3-force@3.0.0
  │ │ ├── d3-dispatch@3.0.1 deduped
  │ │ ├── d3-quadtree@3.0.1 deduped
  │ │ └── d3-timer@3.0.1 deduped
  │ ├── d3-format@3.1.0
  │ ├─┬ d3-geo@3.1.0
  │ │ └── d3-array@3.2.4 deduped
  │ ├── d3-hierarchy@3.1.2
  │ ├─┬ d3-interpolate@3.0.1
  │ │ └── d3-color@3.1.0 deduped
  │ ├── d3-path@3.1.0
  │ ├── d3-polygon@3.0.1
  │ ├── d3-quadtree@3.0.1
  │ ├── d3-random@3.0.1
  │ ├─┬ d3-scale-chromatic@3.0.0
  │ │ ├── d3-color@3.1.0 deduped
  │ │ └── d3-interpolate@3.0.1 deduped
  │ ├─┬ d3-scale@4.0.2
  │ │ ├── d3-array@3.2.4 deduped
  │ │ ├── d3-format@3.1.0 deduped
  │ │ ├── d3-interpolate@3.0.1 deduped
  │ │ ├── d3-time-format@4.1.0 deduped
  │ │ └── d3-time@3.1.0 deduped
  │ ├── d3-selection@3.0.0
  │ ├─┬ d3-shape@3.2.0
  │ │ └── d3-path@3.1.0 deduped
  │ ├─┬ d3-time-format@4.1.0
  │ │ └── d3-time@3.1.0 deduped
  │ ├─┬ d3-time@3.1.0
  │ │ └── d3-array@3.2.4 deduped
  │ ├── d3-timer@3.0.1
  │ ├─┬ d3-transition@3.0.1
  │ │ ├── d3-color@3.1.0 deduped
  │ │ ├── d3-dispatch@3.0.1 deduped
  │ │ ├── d3-ease@3.0.1 deduped
  │ │ ├── d3-interpolate@3.0.1 deduped
  │ │ ├── d3-selection@3.0.0 deduped
  │ │ └── d3-timer@3.0.1 deduped
  │ └─┬ d3-zoom@3.0.0
  │   ├── d3-dispatch@3.0.1 deduped
  │   ├── d3-drag@3.0.0 deduped
  │   ├── d3-interpolate@3.0.1 deduped
  │   ├── d3-selection@3.0.0 deduped
  │   └── d3-transition@3.0.1 deduped
#995531
1600×24001.25Mb
>

Trying to simulate three spatial dimensions on a flat screen will always be imperfect and make it difficult to read the data. However, it is very easy to graph three different data dimensions in D3. You use the horizontal and vertical layouts for two of your data variables, and then size, shape, color or shading for your third variable.

>

If all three of your data variables are best represented by continuous numbers, then your best approach is to use a bubble-scatterplot, where your three display dimensions are horizontal positions, vertical position, and bubble size.


isn't it lame to omit third dimension just because "it's hard to display"

I'm actually thinking about 4D graphs, probably people already tried to display that too

#996209
1560×876115.61Kb

why is she so fat?

baby?

#996319
1500×2251440.95Kb

>>996179

not very realistic

when something doesn't work, I don't hold my head with hands (implying it's my issue), I swear like crazy to the screen, implying another stupid engineer made that happen to me

#996469

made query which runs for 600ms


and I was thinking it's too fast

#996517

>>996469

queries suddenly become scary when theres a million row dataset

#996518

>>996517

I just made it too stupid

run complex regexp on every entry instead of running it once for all items (realized it later)

#996520

even simple context menus you have to build from absoluut scratch on the web i guess there was a reason peope use those component libraries from the start

#996673

so uh if you have a component with bg-blue-500 and then use the component and add some class like bg-red-500 so it becomes class="bg-blue-500 bg-red-500" theres no logic to which rule is used

so you cant override things

it's over

#996677

>>996673

yeah, you have to use something like


<div classList={{ "bg-red-500": p.error, "bg-blue-500": !p.error }}>
#996683

>>996680

yeah, I'm just thinking about that problem


the inheritance is very hard and brittle, because you can change some small thing in child component and break parent component which depended on something


better is to copy-past the full component and change it where you use it

or only accept props which can't break thing later


copy-paste isn't ideal but safer


well, they discovered it in Java decades ago, basic design patterns like https://en.wikipedia.org/wiki/Composition_over_inheritance

#996685

>>996683

>

small thing in child component and break parent component

err, I meant change base component and break child component

#996686

>>996683

css always prefers the latter rule so I thought classes would be the same

but no now I don't feel so good

no quick fixing components in certain context, gotta bake some gigalogic into the component itself to take care of those situations

#996688

>>996686

>

so I thought classes would be the same

they are. you just have no control in which order tailwind put whose bg-red and bg-blue classes in the CSS it emitted


or maybe you can, just look at how it does it but anyway, not a good idea to depend on it


>

bake some gigalogic into the component

yeah, oh copy-paste & change your private copy without the risk of breaking anything

the downside is that those implementations may drift and you won't have consistent look across web app


but I dunno how to do it better anyway, just unsolvable problem of software development in general

#996689

>>996688

>

they are

I mean classes in the CSS, not classes in HTML tag

#996723
850×100662.33Kb

https://www.youtube.com/watch?v=NrQkdDVupQE

I lost it on the black woman part


anyway, after using it for a while, it doesn't see too good. just a bit faster autocomplete

chatting helps only for simplest tasks and autocompletion mostly works for simple repetitive tasks


maybe I'm using it wrong, but often it's faster for me to type what I wanted myself, when wait few hundred ms while it autocompletes what it thinks would fit. because it might autocomplete not the thing I wanted and I actually lost 300ms waiting (I know I can type instead of waiting, but something inside me wants to let it think instead of abusing it with keystrokes or maybe because if you type too much it might not autocomplete at all. need more practice)


also doesn't help much with tinkering with existing code because if you edit inside string it seems to ignore that. only if you adding something at the end of the line


tldr; product is delivering 20% of what marketing promised, as usual

#996799
270×1106.34Kb

>>978550

hopefully i'm still measuring the right thing but

omo

#996802

>>996799

what model are you using?

are you tagging kpop photos?

#997119

>>997118

>

A selfie is a self-portrait photograph

>

A self-portrait is a portrait of an artist made by themselves

=> no

#997126

>>997124

either

1. I have to put too much component-specific logic into the unrelated component

2. Or my component won't properly react to the state changes made by other components


my abstractions suck

dunno how to separate components and still make them behave nicely together


just wanting to do it beautiful

#997129

>>997126

ok, I think I cracked it my mutating the state in other component instead of immutable shallow replace

hackish but works

I think that's a bit better than putting unrelated logic

#997130

things are a bit complicated when you deal with declarative reactivity

have to always remember to make everything declarative otherwise things might go out of sync


on the other hand, imperative UI is a fucking unmanageable mess which is much worse


maybe you can combine by using mostly reactivity but sometimes imperative approach?

#997148
4096×27311.39Mb

some people online seem to swear by a single application or api as the only interface to the database.

makes me curious how I might make that work. currently all pieces of archiver (twatter, tistory, youtube etc) are their own programs that directly access the database


if i had a single database api all the archivers would have to work via post requests to the database owning app, but if i did manage all that i could then try using sqlite instead of postgres. that might be fun


hopefully rubber duck has some ideas about this while i try to remember how my code works, might be the dumbest refactor ever

#997154

>>997148

>

currently all pieces of archiver (twatter, tistory, youtube etc) are their own programs that directly access the database

so you have copy of database wrapper objects/queries in every app? sounds not very effective


you may make small private library which all those apps use which abstracts the DB access

or just merge all those sub-apps into a single project, optionally accessed by different CLI utils. or maybe just a single one like ./app subapp1 --args, ./app subapp2 --args


abstracting DB access with HTTP API may work, but dunno, sounds more complicated than shared DB code

#997156

>>997148

>

i could then try using sqlite instead of postgres

sqlite is only for small databases

you had problem with query performance, you probably have more data than what's suitable for sqlite


may try nosql like mongodb if you don't care about normalization/transactions much

which may be true for simple archiving apps which just store metadata


people tend to hate mongo but SQL feels meh every time I use it, so much useless dancing around the data

and the DB daemon itself is too complex, requires monitoring, tuning, etc

but seems inevitable for reliable data storage, mongo and other nosql trash isn't there

#997169

lowendtalk is one of the shittiest forums I've seen ever

worse than facebook

yet many people share valuable info there

#997336
>

This is a consequence of how DisplayPort is designed, so most DisplayPort monitors will behave this way. If you don't want this behavior, don't use DisplayPort.

>

DVI and HDMI provide a small 5 V power line from the source to the display. This provides a minimal amount of power to the display so that its identification data can be read. Since this power is provided by the source device, it does not depend on the power or state of the display. Even if the display is completely unplugged from the wall, the source will be able to read it and see that a display is attached.


but HDMI doesn't work well in macOS either

#997377

I bought mouse with bluetooth connection because I didn't like the dongle

but then I also bought monitor with usb-c and hub so I can connect mouse's dongle there and have just one cable


what do

maybe return the mouse? but I hate returns

#997402
1000×1000112.02Kb

also bought that thing

spending too much money on electronics this month

#997608
1440×1800295.02Kb

>>997324

macos is dumbo

just to use it remotely on a 4k monitor you need to attach it into another 4k monitor or download some rando's binary from github that creates a virtual display that has a proper resolution

>>997377

mouse usb dongles are so comfy, i was looking at usb dongled headphones earlier to avoid bluetooth memes but they all seem to include a charger so they are too big. i'll have to keep using the bluetooth headphones and bare the magical times every week or so when they stop working for absolutely no reason for half an hour

i've even moved in the middle so it can't be some electromagnetic mystery device my neighbor uses at random times

#997616
2623×1475305.73Kb

>>997608

>

to avoid bluetooth memes

well dongles also use 2.4 GHz, just with a proprietary protocol instead of bluetooth. but maybe that gives them a bit of freedom to implement the connection better. or maybe worse, how can we be sure they tested it in all possible scenarios like when you have many 2.4 GHz wifi AP around


I remember reading reports where people were having issues with multiple Logitech dongles conflicting with each other or something like that...


also I'm not sure the dongle will work OK if I connect it to the pic. I mean what if it's metal parts inside the monitor so the signal won't go through (I had that issue with bluetooth adapter connected to the back of my PC before)


well, can just test, the only problem is that I will receive bluetooth mouse sooner than the monitor so less time to return the mouse if the dongle + monitor solution works also the mouse is nice, can't choose between being cheep and playing with a new toy. welp

#997618
1440×1800313.44Kb

>>997616

i have that monitor except with ergo stand logitech maus dongle at the back of this thin monitor works extremely normally even if i go 10 meters away (microwaves go baunsi bauns https://www.youtube.com/watch?v=QqggrSpICM0)


i mostly like dongles since you can switch the computer with the click of a kvm, but avoiding all the bluetooth issues makes them amazing. devices with wireless receiver have existed for a long time, maybe they are just more polished

#997620

>>997618

>

i have that monitor except with ergo stand

which model?


>

microwaves go baunsi bauns

dunno, my old bluetooth adapter definitely couldn't go through the metal PC case. maybe because of very low powered transmitter tho, logi dongle could be better (because wave loses energy when it's reflecting, compared to when you have the transmitter in a line of sight).


but I believe you, logi adapter will probably work well. welp, should I return the mouse then


>

but avoiding all the bluetooth issues makes them amazing

dunno, bluetooth works mostly fine for me. at least much better than occupying port of the notebook with a dongle. people use usb docks but I don't trust those things to carry signal to my monitor, it doesn't work well even if I connect it directly... or maybe if it's a very expensive thunderbolt hub then should be fine...

#997622
1000×1300258.65Kb
>

thunderbolt

>

thunder thighs

correlation?

#997623
1440×1800236.18Kb

>>997620

lg ergo 32"

i use a regular ol' usb 2.0 switcharooni as my kvm for mouse mic and keyboard, these bad boys don't need some mega bandwidth thingymagic


just tested the dongled directly in the back since you wanted to kno

#997626

>>997623

LG 32UN880-B?

I thought about buying it too

but 32" with same 4K has less PPI also I have another 27" so decided should be ok with 2x27


also I didn't read any reviews or anything. just saw the BF deal, checked that it's relatively ok and just bought. compulsive purchase


>

these bad boys don't need some mega bandwidth thingymagic

I mean when you carry several display signals (4K@60 requires 12.5Gbps), the power and also few USB2.0/3.0 devices such as dongles/external disks through the single cable


if you connect display and dongle separately then you don't need dock


>

as my kvm for mouse mic and keyboard

why do you need that? Linux + Windows?

#997631
1280×720685.87Kb00:03

oh no, I checked, I would have to pay for delivery cost for returning the mouse

which will cost me 30% of the price of the mouse


what do, it would be easier just to keep it

#997633

>>997631

oh, I found another way

I can just don't receive it and it should be sent back without additional cost


I think I will try it

#997643
3072×40962.09Mb

>>997626

i need scaling even on 32" blind nim at your service

>>997631

maybe if you have use for it, i just use desktop so no need for dongle less devices

#997682

>>997643

>

so no need for dongle less devices

I have a need for free ports because it sucks to disconnect everything every time I want to work in different place

so monitor's dock should help me with that, don't need bluetooth mouse. and I quite like my current mouse anyway (g305)


but having the new cool mouse would be great too, you don't need to replace batteries there because it has battery. and it also has horizontal scroll wheel which is useful sometimes. not sure yet

#997740
2731×40961.47Mb

>>997154

yeah i share code currently

>>997156

sqlite would be fine for me at millions of rows scale

the performance problems were due to me dumb ass aggregating related data for each row while seeking instead of after it

>

nosql

i'm not that crazy sql database integrity checks save me from a lot of trouble

some datalog database might be interesting tho if it existed but not that massive java daemon called datomic

>

requires monitoring, tuning, etc

another reason sqlite would be nice but the rewrite would be massive so i'll just continue dreaming for now

#997881
1580×89668.16Kb

>>997880

here is more visible


but probably just lack of data

#997883
1580×820126.09Kb

>>997881

not because of height/age correlation btw

#997931
720×12801.42Mb00:14

>>997402

spent few hours setting this up


I planned to put secondary monitor on the right, but power cable isn't long enough because plug is on the left. have to put it on the left


also seems like main horizontal + secondary vertical would be better layout, because 2x27" horizontal require too much neck work


but if secondary is vertical then power cord may be enough to put it on the right. or many use cord extension. dunno


anyway seems like no much difference if secondary on the right vs on the left


also the new moni setup doesn't play well with my deck lamp had to unmount it. I use it rarely but sometimes it's useful

#997932
1280×72048.39Kb
1280×72047.47Kb

seems like people put secondary on the left


maybe because main = strong hand = right

secondary = left

#997935

people use notebooks in clamshell mode, have to buy stand, keyboard, can't use internal panel, can't use speakers, can't use fingerprint button, have to setup brightness/volume buttons with new keyboard


why would you do that only to avoid neck pain? maybe it's better to train neck muscles

#997936

https://www.youtube.com/watch?v=Dvt7ZPL15xU

Ning looks great in vertical orientation


but vertical monitor seems not very useful

once you open some page which is supposed to be watched in landscape mode, the UX become shitty

even if you have main monitor in landscape, seems like it's better to have secondary in landscape too


should I buy mx keys mini + laptop stand tho?

#997971
3024×40321.81Mb

>>997936

I couldn't get used to double monitors so i just go with the one at least i can pivot for the ningies when necessary


maybe if there was a fun litl monitor that was as tall pivoted as my main monitor landscaped with the same ppi and i managed to avoid the cable hassle somehow

#997973
608×10802.92Mb00:10인싸들의 계단 올라가는 법 <https://www.youtube.com/watch?v=UeruzOtlJ8g>

>>997935

just simply buy an apple external keyboard with apple t2 security chip and touch id

#997974

>>997971

>

I couldn't get used to double monitors so i just go with the one

I just wanted 4K monitor because macOS doesn't play well with 1440p monitors (because it can't into sub-pixel rendering). and since I already have one, then it would be 2


also I currently use internal screen + external monitor and it's not always enough to put all windows I want (when developing you often need many windows open). so 3 (2 external + 1 internal) seems like the absolute minimum


>

i can pivot for the ningies when necessary

dunno, feels like a chore to rotate it back and forth even if you have a nice stand. better to have separate portrait monitor if you need it often. but seems like in reality vertical monitors aren't very useful


>

with the same ppi

does it matter? at least macOS doesn't care if different monitors have different PPI


>

i managed to avoid the cable hassle somehow

that thingie >>997402 has a nice niche for cable management


>>997973

too expensive


also, I don't understand why do they use clamshell mode with separate keyboard if you can type on notebook's keyboard and look into external monitor. weird

#997975

>>997974

>

too expensive

hm, just checked, actually just few dollars more than mx mini


logitech nim is too greedy


but there's also MX Keys Mini Combo with mx anywhere mouse


uh huh, shouldn't have looked into the keyboard/mouse stuff

I was fine with my current mouse, now I want all the new shiny stuff

#997977

>>997976

weird video, why do they compare thunderbolt 4 (which can deliver 40Gbps) with some random USB-C which is probably just 10Gbps

but definitely you can find cheaper than Apple's on aliexpress. but it wouldn't be very cheap either


also I just said that logitech accessories are pretty expensive (at least the hyped one like mx mini/mx master series). I didn't say that apple's accessories are cheap

#997980

>>997976

also there's error in the video. when they talk about amazon's cable they say there's no IC inside and that it's fine to charge stuff


but to charge over USB-C you have to have controller


stupid video

#997982

>>997981

>

the webm


>

this would solve half my problems with blutooth

what? the dongle?

#997985

>>997983

oh yeah, some bluetooth devices support multiple paired devices


omo Juri!

#997986

>>997984

also you didn't answer why do you need multiple devices