Linux @lemmy.ml

2w ago

What's a unique customization on your Linux machine you think no one else has?

Just wondering since I know a lot of people quietly use a screen-area-select -> tesseract OCR -> clipboard shortcut.

I separate subjects of interest into different Firefox windows, in different workspaces -- so I have an extension title them and a startup script parse text to ask the compositor to put them in the correct workspace (lets me restart more conveniently).
I have automatically-set different-orientation wallpapers for using my 2-in-1 depending on whether I use it in portrait or landscape (kind of just for looks, but I don't think if anyone else adds a wallpaper change to their screen rotation keybind).

119 comments

I use my DE mostly as it comes, that's got to be unique in this community
- Some people use plasma because they like how configurable it is. I do like that, but I'm also drawn to it because of its great defaults.
  The main ways I change it are setting my background (on my work activity I have it selecting from various company related backgrounds while on my personal activity it uses a selection of my favourites of my own photos) and adjusting the bottom panel.
  
  Funny you should say that, I always felt like the defaults are really bad.
- Probably, I have about 20 extensions for GNOME and have tweaked right about every setting and keybind.
  
  I just like the extension that lets me swap audio devices without delving into settings

When I press Super + PrtSc, a bash script performs the following:
Takes a screenshot of the entire desktop (import -window root) and saves it as ~/screenshot.png..
Analyzes the screenshot to calculate the "mean brightness" value of the image. It converts the image to grayscale and determines the average pixel brightness (a value between 0 and 1, where 0 is black and 1 is white).
Checks if the image is dark by comparing the mean brightness to a threshold of 0.2. If the mean brightness is less than 0.2 (i.e., the image is very dark), it applies a negative filter to the image (convert -negate), effectively inverting the colors (black becomes white and vice versa).
Sends the image to a printer (lp command) named MF741C-743C for printing.
- an actual print screen, finally
  
  A kind of ‘super’ print screen, in fact.
- why?
  
  Honestly I print out anything my little kiddo does at school on his Chromebook, and some stuff has black backgrounds. I got tired of wasting toner so I made a script that would print a negative screenshot if it's a dark image. One keystroke and I get what I want

I use KDE's defaults.
- That's sick man! Get some help!
- I use Gnome defaults.
  
  With Gnome you have no other choice.

I have two mice, one for either hand, and use xinput to flip the buttons on JUST the left one. It's actually one of the main things keeping me from moving to Wayland, which doesn't seem to have the same configuration features
- LOL I've never seen that before.
  Do you use them both at the same time? Or do you switch between them rapidly? (Maybe you could make a taskbar button-toggle if it's the latter!)
- there are both configurable mice that let you swap mouse buttons (in the worst case in a windows virtualbox with usb passthrough) or mice that are leftie right out the get-go. those would allow you to use wayland, assuming you can afford to and want to get a new mouse.

I have an old gamer keyboard with extra programmable keys on the side, which I use for cut, copy, paste, close tab, close window, etc. Logitech provides drivers/software for Windows & Mac only.
To make it work I have a custom monkey-patched USB driver that I compiled from source, some weird daemon that interacts with the driver and some shell scripts on top of that. I'm not sure how but it works thanks to a 9 year old youtube video made by a guy from eastern europe somewhere.
- Awesome...
  Care to share the video/code? ~~I actually have something similar (Corsair Scimitar's macro customizer doesn't work on Linux~~
  As I was writing this I found a project that deals with Corsair MMO mice on Linux so now I will be going on an egg hunt.
  
  video https://www.youtube.com/watch?v=8gAT-BbyOWw
  code https://github.com/Leproide/Linux-G15-Daemon-Logitech-G110-
  I'm pretty sure it will only work with a handful of old Logitech keyboards.
  When I eventually upgrade my OS and can't compile the stack for some reason, I've got a Sun Type-7 waiting in the wings.
- I do something similar.
  I have a V4N4G0N that I use the top row (half the normal number row on a full sized board) for switching workspace or switching apps to another workspace, and doing other stuff like copy and paste on different layers for the keyboard.
  As its QMK (via VIAL) I have set all that up directly on the keyboard so its portable to any other PC I want to use. I have eight of these, mix of alu, acrylic and 3D printed, that I can choose from, all sharing the same map. I don't like using anything else now as its become integral to my normal workflow.

I suspect my habit of having an alias userctl="systemctl --user" is slightly unusual, as is running Firefox, Steam, and some other graphical programs as systemd units is somewhat unusual (e.g. mod4-enter runs systemd-run --user alacritty)
But what I'm actually pretty sure is unique is my keyboard layout. I taught myself dvorak a summer some decades ago, but the norwegian dvorak layout has some annoyances, so I've made some tweaks. Used to be a Xmodmap file, but with the switch to wayland I turned it into a file in /usr/share/X11/xkb/symbols/.
Part of what I did to teach myself dvorak and touch-typing at the same time was randomize the placement of the keycaps too. It has a side effect of being a kind of security by obscurity layer: I type quickly and confidently, but others who want to use my machines have an "uhh …" reaction.
- I have been using the same userctl alias.
- I didn't know about the systemd-run command. Do you use it to save the command log? I created a script conveniently named x which opens a file in a default app, in the background, so I can still use the terminal. But then I had the problem with handling logs and this sounds like a perfect solution. Gonna try it today.
  As for the alias, I wanted to create a pacman-like interface for systemctl, so the commands would be much shorter, but never finished it. For example, sctl -Eun unit would be equal to sysyemctl enable --user --now unit
  
  The logs are handled, but I mostly use it for command separation and control, including killing unruly child processes.

I made a user for my partner
- I also have a user for your partner
  
  Haha!
  Take my poor man's gold
  🏅

While I doubt the concept is unique, the script is: a keyboard shortcut will check the clipboard for a YouTube link and then show launcher options for mpv or yt-dlp, including launch arguments for lower quality format and audio only. It launches that in a terminal for easier handling when yt-dlp doesn't work properly (much more common if using proxies, but also if a video is age-restricted or deleted).
So when I see a yt link here, I can just copy it, keyboard shortcut and then it's playing in my local video player.
edit: here's the script. It assumes xsel (clipboard access), rofi (menu creator), gnome-terminal (terminal) and notify-send (system notification on failure) are installed and working, you'll need to replace any which don't match your system. My DE just runs it in bash when the shortcut is entered.
- Uh I would be interested in that actually! Nowadays Youtube generates lots of problems with freetube due to their cookie bullshit and I feel with mpv(yt-dlp) in cli I at least have the option to see whats going on.
  
  Now added to my comment :)
  
  Same
- Uh this sounds awesome, care to share?
  
  Now added to my comment :)

Machined badge reading "Built Not Bought".
My dad used to put them on the cars he built.
- My dad used to put them on the cars he built.
  That's pretty rad.
  
  He was a rad guy.

I use compose key sequences to save time writing out long email addresses. For example, I have something like this in my ~/.XCompose:
undefined

<Multi_key> <b> <o> <s> <at>: "myangryboss@company.com" # Email of my very angry boss
So I can just type Compose (right alt on my system), bos@ and get his email address. Less error prone than typing out emails manually.
I'm probably not the only one to use compose strings as a replacement to a text expander, but I don't know anyone else who does this.

Maybe a bit plain since I'm only at mediocre level in my Linux journey, but I use my favorite fonts for Kitty. Recursive Mono Linear and then for italics and comments in neovim I use Recursive Mono Casual Italic.
Recursive Linear is so tidy and neat, with just the lightest touch of personality. And Casual keeps that style but tweaks it just ever so slightly to a more comic. And they have sans versions of both as well for everything else.
I also made my own Starship prompt to match my desktop. It runs an easily reconfigurable color palette and uses color coded chevrons to denote different git statuses.

I'm one of at most a handful people in the world with a full disk encrypted Steam Deck and unlocking using the touchscreen.
Until someone implements https://github.com/ublue-os/bazzite/issues/464 in Bazzite.

Small thing, but I really like it: I have ~/autoclean_tmp directory on most of the hosts I use as a desktop. Then on crontab I have a find-command which automatically deletes files which are 7 days or older. I can throw stuff I download from the internet and copy from other hosts, random text files when setting up new stuff and so on in there and they just vanish after a while.
- I have the same type of thing. An alias that creates a tempdir that is based on the date, then cd's into it. Then a cron job that finds dirs that are older then N days old and deletes them. I use these for most of my scratch work. Having several days to look back at what you did and know when you did it is so nice.

I boot on a custom EFI app to control my dualboot (instead of systemd-boot or grub) that asks a service on my proxmox server which OS I'm supposed to boot.
Overkill, but it allows me to control my dual-boot without a keyboard in my computer (because it's a Bluetooth keyboard so I can't really use it in grub anyway)
- A custom EFI app? Is that like a handrolled Unified Kernel Image with some Proxmox-specific addons in it? How'd you make it?
  
  No, it's a EFI app I developed in Rust that does a query over multicast UDP and uses the result to select which EFI app (Windows bootloaded (yeah I know...) Or systemd-boot to start Arch)
  There's nothing related to proxmox itself, it's just there that I host my LXC with the service that responds to the quey.

the ability to use two Bluetooth dongles simultaneously, each for one device. try that on Microsoft's clown os and see how pressing the gamepad triggers makes the bluetooth headphones chop up the sound 😂

In all my servers I still have a cron->make routine running. It's a hold-over from 20 years ago and the state of IaC back then, and it's made its way onto every server I manage because it is simple and effective.
And it still does its job. 8 major RHEL releases later, and the thing it needs to do, it does.
Lennart would build 3 new daemons and link them all into dbus, I'm sure.

I have Syncthing set up to copy save data between my pc and steam deck, but not just for emulator stuff: its got my entire modded minecraft directory and my balatro modloader nn there too.
- Syncthing is great and incredibly easy to use. I have mine set to sync my Obsidian notes so I don't have to pay for the official service.
  I have tried multiple different open source note apps that offer free local sync, but I can't find anything I like. It frustrates me because I love open source.
  
  I use the same setup with Syncthing and Obsidian. The git plugin sometimes gets confused, but nothing I can't untangle. I also use Syncthing for pictures off my phone, and ebooks onto it.
  Actually, I think I do have a setup that might qualify as unusual: I use the scheduled backup feature of Podcast Addict to get a listing of listened podcast episodes, and then I inject them into my Obsidian notes.
  
  Take a look at the Self-Hosted LiveSync plugin for Obsidian. Requires some self hosting for a sync server, but it is damn flawless. I have my phone, desktop, laptop, and work laptop, all syncing through it. Syncs live too, so you can even see me typing on one device from another

Does stuff I wrote myself count?
Apache server that has a bunch of webpages that are all configured by simple JSON files and loaded by PHP. The pages have buttons on them which when pressed enter macros. So I push "Deploy Landing Gear" and Shift+alt+F8 or some obscure as fuck combination no one would ever use normally gets pressed and the game can be set to use that keybind. Most of it is for simple immediate key presses but also made a few for macros as well.
The HTML/PHP that runs the show is a grand total of 2018 bytes, including comments. Plus a fairly bloated 2444 byte CSS file that includes some button colour options that I never use now because I decided they look ugly. Should update some of the background images though, my sheet steel Faulcon DeLacy logo looks a bit basic.

My applications menu icon (or the “start” menu for the philistines) is a 🐢.
- based

I have stickers on it, some of them hand drawn by my daughters.

The text editor shortcut on my taskbar runs a sort of autosave script in ~/.drafts. I wanted my text editor to function more like the one on my phone so I can just jot down random thoughts without going through the whole ritual of naming and saving. It creates YYYYMMDD_text in ~/.drafts (or YYYYMMDD_text_1 etc. if it already exists) and launches Pluma, which I also have configured to autosave every 10 minutes.
The other thing extends beyond Linux itself a bit. I like to joke that I have the most secure NT 4 / Windows 95 lookalike ever put together. Aside from the encrypted and hardened Debian base (/boot is also encrypted), I was in part inspired by Apple's parts pairing (yikes!). So my coreboot is configured to only accept my boot disk. If it's swapped out or missing, or if I want to boot something else, it will ask for a password. In the unlikely event my machine gets stolen, the thief must at a minimum reflash the BIOS or replace the motherboard to make it useful again. Idk, it amuses me every time I think about it.

I created my own openSUSE splash screen for KDE because I felt all the existing ones were a bit amateur and I wanted something professional looking. I haven’t published it because I can’t be bothered creating an account. It only took about 15 minutes because I chopped up another one which had clearly chopped up another one.

I am indecisive when it comes to wallpapers so I have a script somewhere which accepts tag-words as arguments and then scrapes wallhaven.cc for those words at the resolution of my setup and picks one that contains those words at random before downloading it to my wallpapers folder and setting it as my wallpaper image.
So for example, you could just know you want something blue so you would run wallpaper blue and it just grabs one and sets it. You could get a wallpaper of the sky, of a blue car, of the ocean, whatever happens to be a wallpaper that met the criteria of the word/s supplied.
- Risky business considering there's always some horny anime crap mixed in on Wallhaven.
  Filters and tags only help so much since lots of it either has poor tags or no tags at all.
  
  There is a toggle for SFW/Sketchy which in my experience has worked pretty well in avoiding such things, but you are probably right it does not catch everything.
  If such a thing happened, I would just re-run the same command to update to a different one though. I guess I generally just make sure no one is in the room when it runs haha.
- KDE actually has a plugin that does just that, I use it currently to rotate a fantasy illustration as my wallpaper every hour from that site.
  
  Oh neat!
  My script is for gnome, but I wonder if there us an equivalent gnome extension in existence as well.

I have bash scripts light and dark that make dbus calls to set my global theme to light or dark mode. I switch between them regularly, and opening system settings and pressing a button is too inconvenient.
Your first one sounds similar to me though - I use activity-aware Firefox to separate my personal and work accounts on my personal and work plasma activities.

I type "power..." into my cli and press tab+enter to shutdown my computer. Same for reboot... 😆
- I do the same except I only type "pow" :P
- Why though?
  
  Because its fast and easy? And also it works regardless of what DE/WM I am using.

My keyboard automatically change the keys depending of the app I'm using: closing a tab in the terminal or closing a tab inlthe browser are always the same key.
https://git.chimrod.com/smartcropad.git/about/

I'm using XFCE with Compiz, and since I have two monitors I have a 3D octagon instead of a 3D cube desktop.

Custom cowsay written in Rust that pulls German song lyrics from my favorite band from a text file?

I spilled a glass of scrumpy on the keyboard and a, s, and d no longer work. So I have to use a keyboard with it.
- So you have to use a keyboard with your keyboard...
  
  With my laptop.

I'm pretty sure no one else has my shell script that takes a picture, uses imagemagick to copy a scaled down version of it to a special folder, and then build a string that allows me to just middle click paste the image into Rednotebook so it appears correctly.
- ooh I should do that for Obsidian instead of having an enormous directory of Pasted Image 202302050124300845012.pngs. =◡=
  
  I use https://github.com/trganda/obsidian-attachment-management to automatically rename and move screenshots, in conjunction with https://github.com/Mara-Li/obsidian-explorer-hider to hide them. It makes pasting screenshots organized, yet completely transparent.

Definitely not nobody but statistically VERY FEW people will have this combination:
pop!os (fight me!)
script that limits accumulator charge to 80% on asus laptop
script that turns on vpn if out of home and kicks off a backup if at home (through wifi ssid)
Edit: nice try to fingerprint me, big tech. You succeeded! /j
- Triangulating your location. Are you... in the Milky Way Galaxy?
  (Thanks for reminding me to limit accumulator charge)
  
  Yeah, I have a script that toggles my Dell XPS between full charge and 80%, as I’m usually on mains and only need full charge occasionally.

I have a meshtastic script that runs once a day that sends a weather report for our local area at 6:00 am. It was based off a script that some awesome person did. I also have a script that once a week sends out ham/meshtastic events to all local people. Its worked out pretty well.

CTRL+SHIFT+L to sync my room lights to the screen using huenicorn. Plan on hooking up openrgb as well when I can be bothered to write a script.

I use this app (webapps is the name I think) to make apps for YouTube, Mubi and TorrentLeech and I have then pinned on the task bar and use them as apps instead of webpages. This is in my hometheater pc

On my desktop, I wrote a Python script that pulls a random Star Trek: The Next Generation or Deep Space Nine script from a folder and prints it in STDOUT. I use this in the XScreenSaver Text Manipulation > Program option to turn Star Trek into a screen saver.
Currently, I use it with the Apple II screensaver, but in its original incarnation, I used the Star Wars intro screensaver. 😈

wait how does your clipboard shortcut work op? that sounds nifty!
- I think I mentioned it but here it is again in case the comment didnt federate
  
  why do you even need a temporary file?
  bash
  
  $ slurp | grim -g - - | tesseract stdin stdout -l eng+kor+jpn | wl-copy -t 'text/plain'

I've got a RPI running a full-screen 'kiosk' view from homeassitant that turns an external display on/off based on a motion sensor.
So basically it's showing current temperatures, thermostat control, etc. but I have the display turn off after X minutes of no movement and turn on when there has been movement so it's only on when you're in the room.
- esphome or custom?
  
  Much simpler than that - The motion sensors are zigbee and integrated with HomeAssistant. I have a HA automation that sends a REST call to a webservice I wrote on the PI that then just needs to write 1 or 0 to /sys/class/backlight/rpi_backlight/bl_power.
- I'd love to see your implementation specs, code, pr any other technical details you'd like to share. I'm setting up home assistant and one of the things I want it to do is replace the functions of my thermostat and add some additional details.
  I used to have a Nest Thermostat, but my furnace needed to be replaced a couple of months back and I got a Mitsubishi heat pump, but their thermostat sucks, and it isnt compatible with Nest because it's all wireless. I installed the WiFi add-on to the furnace so I can use the app, too, but it also sucks pretty bad. Plus I miss the functionality of it turning down the heat when I'm away to save money and turning it back on before I get home.
  So I'm planning to implement my own solution and documenting and open sourcing everything. But it's going to be several months before I get to doing it due to other more urgent projects. So, I'm looking at everything available. I definitely will be setting up a small display to replace the thermostat and having motion detectors to turn on the display when you approach it to see the temperature and such and to supplement the home/away detection.
  Anyway, I would love to see your implementation to see how you did this piece of it.
- I have a similar display in my kitchen. It's in portrait mode and has time (my timezone and others), weather (hourly and daily), and dynamic popups for weather alerts in the top 1/3. It has a spot for dynamic content below that that shows things like time remaining for my espresso machine to heat up and the temperature of my ember mug if I'm using it. The bottom half of the screen flips every 15 seconds between calenders for my partner and I, and local scheduled transit times and live train times with a map of current train positions.

I use many KDE activities all mapped to a single hotkey. Meta+H, Meta+J, Meta+K, then L, Y, U, G.
I set my browser and maybe one other as sticky to show on all. I also have specific desktop picture for all of them.
On top of that I have a startup command that opens all applications I use for work. Each application is configured to open in a certain activity.
The end result is that instead of doing Alt-Tab or looking for the window I do Meta+Key and it's there in front of my eyes with focus.

I have scripts set up to switch between my desk setup and my home theater setup that swap monitor configurations with wlrandr and default audio devices in wireplumber. These scripts are triggered with the "Netflix" button on my Nvidia Shield remote via Home Assistant and SSH. Simultaneously on Home Assistant power to the peripherals on my desk is toggled, the TV input is toggled between the Nvidia Shield and the PC, my AV receiver settings are toggled, and if the PC was asleep, it's turned on with a WoL magic packet.

Coming from Windows, I set up KDE's Spectacle to open with Super + Shift + S in Area Select Mode and save and copy to clipboard on click release
Maybe not as unique but kinda neat I think
- I do this too. just a very slick hot key combo imo
- Not unique, but we are now kindred (I did the same <:)

I have a custom script, which changes the fan profile (in my case between two thinkfan config files) depending whether the dock is connected or not. That one gets triggered whenever it switches the power source (AC or BAT0). (AC gets plugged in -> script starts -> check if dock is connected -> if connected run different profile)
It's janky but very helpful when it works :D

I've got basically the bspwm workflow, but on KDE.
So, bspwm has tiling of windows and doesn't want you to minimize (nowadays, it actually has a minimize-feature, but back when I last used it, it didn't). As a result, if a window is open, it is visible on some workspace. If you want to hide windows, put them on a different workspace.
I like that workflow, because while it probably seems complex when you first hear about it, it actually simplifies things. When you're looking for a window, you don't have to check all the workspaces and minimized windows and behind other windows.
KDE adds to that, in that I can have a workspace overview in my panel, so where I can see all workspaces with the windows that are visible on them (which with this workflow is all windows on that workspace). I like to call it my minimap.
It makes the workflow a lot easier to use, but it also allows me to group workspaces by location. So, if I'm working on a topic, I often have a Firefox window on one workspace, my text editor on the workspace below and then a terminal on the workspace below that. If I then realize, I need to quickly look up something for a related topic, I'll open up a new Firefox window two workspaces below that (leaving an empty workspace as separator). If I do something completely different, I might leave a whole bunch of empty workspaces in between. Or, well, KDE actually allows grouping workspaces with a feature called "Activities", so I'll often switch Activities.
I find that works a lot better for multi-tasking than the traditional Windows workflow of one window per application, with all kinds of different topics mixed into all kinds of ungrouped windows. If I switch between topics, I just go to the right location on my minimap and I've all the topic-related information in the windows that are there.

I also seperate Firefox on different workspaces, but only manually. How is the extension called? Having it automated would save me some seconds every reboot.
- Sway (and i3) you can assign windows to workspaces based on any property that is available in the swaymsg tree. It can do parital matches, so for example if you wanted your Lemmy firefox window to always start on workspace 3 you could use:
  assign [title="lemmy" app_id="firefox"] workspace number $ws3
  Title can use regex so you can do some pretty neat matching if you need it.
- edit: based on the other commenter I think I might be missing a simpler declarative way to do this. The following will be kept for posterity though
  The main idea is:
  Use Window Titler to add a title. For me, if I want it on workspace 7, I title the window "7". (NOTE: The title will probably appear like [title], see below)
  Make a script that queries the window manager, and then dispatches a movement to the appropriate workspace. In Hyprland that might be hyprctl -j which gives
  bash
  
  ... json blahblah "title": "[7] What's a unique customization on your Linux machine you think no one else has? - tchncs — Mozilla Firefox" ... json blahblah
  but in Sway it might be something similar to using swaymsg. Only titled windows will have the bracket number thing so just regex that part
  Put it in autostart. Because Firefox takes a while to load on my junk machine I sleep for like 30 seconds to a minute before all the titles register.
  
  Thanks a lot!

I use hot mount SATA slots for backup and other media. Not that common on workstations. Sure, common on servers.

"yubi [website name]" in Alt+F2 — asks yubikey for a TOTP code for a website and autotypes it into wherever I've got my focus

my awesome wm config has a lot of customization. We're talking 5+ years of basically re-writing an entire theme, along with behaviours, widgets, and bindings.
- Which WM?
  I am new to themes with gnome and am interested in learning about it in that capacity if you should have any resource material saved!
  
  AwesomeWM

I have a zellij snd micro config for journaling and writing that makes a completely borderless full screen terminal with no decoration whatsoever and narrows the terminal for micro to the upper half of the middle 1/3 of my screen.
It helps me focus and limiting to the upper half and middle 1/3 makes it easier for my eyes. I get distracted easily and this helps keep my editor from being the source of that.

ChatGPT wrote a Python program that does select->Tesseract OCR for me, but it doesn't always work right with two monitors. I'm too stupid to correct it. How have you done yours, what are you using for selecting the area?
- pasting from my keybind config
  bash
  
  # snippet based on end4 dotfiles -- FIXME edge case where a # preexisting tmp.png might be overwritten # English bind = Super+Shift,T,exec,grim -g "$(slurp $SLURP_ARGS)" "tmp.png" && tesseract -l eng "tmp.png" - | wl-copy && rm "tmp.png" # Korean bind = Super+Shift,K,exec,grim -g "$(slurp $SLURP_ARGS)" "tmp.png" && tesseract -l kor "tmp.png" - | wl-copy && rm "tmp.png" # Japanese bind = Super+Shift,J,exec,grim -g "$(slurp $SLURP_ARGS)" "tmp.png" && tesseract -l jpn "tmp.png" - | wl-copy && rm "tmp.png"
  I just pipe grim and slurp (i.e. select part of the Wayland screen then copy) into a temporary png, tesseract it into the clipboard, then delete the temporary png.
  edit: clarified

119 comments