tigervnc: vncserver@:1.service: Deactivated successfully just after start on gentoo

Tigervnc server starts and then stops immedietly. The solution was to go to /etc/X11/Sessions, and rename Xsession file:

cd /etc/X11/Sessions
mv ./Xsession Xsession_bkp
And restart the vncserver.

chroot: failed to run command ‘/bin/bash’: No such file or directory

You can specify different shell, e.g.: sudo chroot /mnt/nixos /bin/sh

xpra: More responsive remote connections

xpra start ssh://server/ --exit-with-children --start-child=terminator --encoding=jpeg --min-speed=100 --min-quality=0 --compress=1 --bandwidth-detection=yes --pixel-depth=16 --auto-refresh-delay=0 --video-scaling=2 --dpi=96
--min-speed — higher is faster. 100 is the maximum, meaning "always encode as fast as possible, never slow down for quality."

--min-quality — lower is faster. 30 means "allow aggressive quality reduction to save bandwidth." Setting it to 0 lets xpra reduce quality even further dynamically

Or run persistent server on the remote server

On the server: xpra start :100 --start-child=xfce4-terminal
On the client:
xpra attach ssh://curtin/:100 --encoding=jpeg --min-speed=100 --min-quality=0 --compress=1 --bandwidth-detection=yes --pixel-depth=16 --auto-refresh-delay=0 --video-scaling=2 --dpi=96

MATLAB on Arch Linux: java.lang.NullPointerException and MATLAB does not start

If MATLAB stopped launching after a recent Arch Linux system update and you see a java.lang.NullPointerException with a stack trace starting at DTSingleClientFrame.mainFrameCreated, likely reason could be a corrupted preferences folder.

You can confirm the issue by checking whether MATLAB works without the desktop:

matlab -nodesktop -nosplash

If that launches fine, the desktop preferences are to blame. The fix is simply to delete MATLAB's local preferences folder:
rm -rf ~/.matlab/R20*/
matlab

MATLAB will recreate the folder fresh on next launch and the desktop should start normally again. No reinstall needed.

picom: Sample confing in gentoo with xfce4 and nouveau dirver

Picom on Nouveau: Which Backend to Use If you're running the open-source nouveau NVIDIA driver and using the picom compositor, you've likely run into rendering glitches or crashes. The culprit is almost always the backend. Picom's GLX backend relies heavily on OpenGL, and nouveau's OpenGL support is incomplete. Stick to the xrender backend instead — it's slower, but stable. Create a config file at

~/.config/picom/picom.conf

with the following content:
backend = "xrender";

Vsync — disable, nouveau's support is unreliable
vsync = false;

Shadows
shadow = true;
shadow-radius = 7;
shadow-opacity = 0.75;
shadow-offset-x = -7;
shadow-offset-y = -7;

Fading
fading = true;
fade-in-step = 0.03;
fade-out-step = 0.03;

Opacity
inactive-opacity = 1.0;
frame-opacity = 1.0;
Then restart picom: pkill picom; picom -b
That's it. You trade some visual effects (blur requires GLX) for a stable desktop. If you ever switch to the proprietary NVIDIA driver, revisit the GLX backend for better performance.

gentoo: Extra packages for xfce4

sudo emerge -av xfce-extra/xfce4-whiskermenu-plugin xfce4-cpugraph-plugin xfce-extra/xfce4-notes-plugin x11-apps/xhost app-admin/system-config-printer xdg-desktop-portal-gtk xfce4-sensors-plugin

arch: keditfiletype missing

yay kde-cli-tools

gwenview: Does not start due to setting rendering to OpenGL

Delete its config file:
rm ~/.config/gwenviewrc

distrobox: Opening Images in Native GIMP from IrfanView Running Under Wine in Distrobox

I use IrfanView via Wine inside a Distrobox container and wanted its external editor feature to open images directly in native GIMP. IrfanView passes the image path in Windows format (Z:\home\home-folder\Pictures\photo.jpg), so a small Python script handles the conversion and launches GIMP. Sounds simple. It wasn't.

Two things needed fixing before GIMP would even start. First, GIMP's sandboxed image loader (glycin) uses bubblewrap, which cannot create user namespaces inside a container — fixed with GLYCIN_DISABLE_SANDBOX=1. Second, any process launched from the script died the moment the script exited, because Wine kills all child processes when its own child (the Python script) returns.

Even after solving those, GIMP crashed every time it tried to open a file, with a GTK3 assertion failure inside GtkFileChooserWidget. The same command worked perfectly from the shell. The environment variables were identical. The difference turned out to be invisible to env: Wine installs a seccomp filter on its processes, and that filter is inherited by every child process across fork and exec. No amount of environment tweaking or session detachment can clear an inherited seccomp filter. GTK's file chooser widget makes a syscall during initialisation that Wine's filter blocks, and GIMP crashes.

The fix is to launch GIMP as a transient systemd user service with systemd-run --user --no-block. A process started this way is owned by systemd, not Wine — clean process state, no inherited seccomp filter, completely outside Wine's process tree.

#!/usr/bin/python
import os, sys, subprocess

def convert_wine_path(p):
    return p.replace('\\', '/').replace('Z:', '')

env = os.environ.copy()
img_file = convert_wine_path(sys.argv[1]) if len(sys.argv) == 2 else None

cmd = [
    'systemd-run', '--user', '--no-block',
    '--setenv=GLYCIN_DISABLE_SANDBOX=1',
    f'--setenv=DISPLAY={env["DISPLAY"]}',
    f'--setenv=HOME={env["HOME"]}',
    f'--setenv=XDG_RUNTIME_DIR={env["XDG_RUNTIME_DIR"]}',
    '/usr/bin/gimp'
]
if img_file:
    cmd.append(img_file)

subprocess.run(cmd)


Save it somewhere on your Linux filesystem, point IrfanView's external editor to the Z: path equivalent, and it works. The same trick applies to any native Linux app you need to launch cleanly from within Wine.

gentoo: There was an error during CUPS operation: "No such file or directory".

This happend for me with gentoo, cups and hplip. Had to install filters:
emerge --ask net-print/cups-filters End then
systemctl restart cups

gentoo and distrobox: OCI runtime exec failed: exec failed: unable to start container process: error starting setns process: exec: already started

I have gentoo host, with systemd, nvidia and docker. When I want to create a distrobox image with systemd in the container, e.g.:
distrobox create --nvidia -i archlinux -n abox -H ~/home-abox --init --additional-packages "systemd"
I was getting: OCI runtime exec failed: exec failed: unable to start container process: error starting setns process: exec: already started
The solution was to use crun container runtime, instead of default runc. To do this

1. Install crun: emerge -av app-containers/crun
2. Create `/etc/docker/daemon.json` with the content: {
"runtimes": {
"crun": {
"path": "/usr/bin/crun"
}
},
"default-runtime": "crun"
}
3. Restart docker sudo systemctl restart docker

bedrock linux: share getnoo's /etc/portage with stratas

For example file managers install in strta, e.g. mc, yazi, will not be able to go /etc/portage of gentoo. To fix that, share portage folder in

Add portage to [global] in /bedrock/etc/bedrock.conf: etc = adjtime, crypttab, default/grub, fstab, group, group+, group-, group.OLD, group.org, gshadow, gshadow+, gshadow-, gshadow.OLD, gshadow.org, hostname, hosts, login.defs, machine-id, modprobe.d/blacklist.conf, passwd, passwd+, passwd-, passwd.OLD, passwd.org, rc.local, resolv.conf, resolvconf/run, shadow, shadow+, shadow-, shadow.OLD, shadow.org, sudoers, portage
Then sudo brl apply

librewolf: dissable adding application for mailto links

about:config
Set: network.protocol-handler.external.mailto to false.

xterm: set font size

Create ~/.Xresources with the content:
! Use a truetype font and size.
xterm*faceName: Monospace
xterm*faceSize: 14
Run the following to take effect: xrdb -merge ~/.Xresources
From: https://askubuntu.com/a/1138738

gentoo: example make.use

# These settings were set by the catalyst build script that automatically
# built this stage.
# Please consult /usr/share/portage/config/make.conf.example for a more
# detailed example.

COMMON_FLAGS="-O2 -pipe -march=native"
CFLAGS="${COMMON_FLAGS}"
CXXFLAGS="${COMMON_FLAGS}"
FCFLAGS="${COMMON_FLAGS}"
FFLAGS="${COMMON_FLAGS}"

# NOTE: This stage was built with the bindist USE flag enabled

# This sets the language of build output to English.
# Please keep this setting intact when reporting bugs.
LC_MESSAGES=C.UTF-8

RUSTFLAGS="${RUSTFLAGS} -C target-cpu=native"

MAKEOPTS="-j4 -l5"

USE="dist-kernel"

# Appending getbinpkg to the list of values within the FEATURES variable
FEATURES="${FEATURES} getbinpkg"
# Require signatures
FEATURES="${FEATURES} binpkg-request-signature"

ACCEPT_LICENSE="*"

EMERGE_DEFAULT_OPTS="--binpkg-respect-use=y"

gentoo: set --binpkg-respect-use in make.use

Add to make.use: EMERGE_DEFAULT_OPTS="--binpkg-respect-use=n

gnome-boxes from flatpak: file selection window does not show

I had that issue on gentoo running systemd with xfce4. Even though xdg-desktop-portal and xdg-desktop-portal-gtk portals were installed and running, gnome-boxes file selection window was not working. The solution was to set gtk portal by default:

create ~/.config/xdg-desktop-portal/portals.conf:

[preferred]
default=gtk
org.freedesktop.impl.portal.FileChooser=gtk


Open WebUI: Add your OpenCode Go subscription

For the connection url use: https://opencode.ai/zen/go/v1 and then your OpenCode Go API key

https://i.imgur.com/KU4yEsj.png

Qwen-Code: Ollama Ignores num_ctx from settings.json

If you're running Ollama behind an OpenAI-compatible endpoint and setting num_ctx in your client's config, you may notice Ollama still loads the model with only 32k context. This is a silent failure — no errors, just ignored parameters.

Why It Happens

Ollama's /v1 OpenAI-compatible layer only forwards standard OpenAI parameters. Native Ollama options like num_ctx need to be passed inside an options: {} object in the raw request body. Most OpenAI-compatible clients don't do this, so the parameter is quietly dropped and Ollama falls back to the model's default context size.

The Fix

Bake num_ctx directly into the model via a custom Modelfile. This sidesteps the client entirely.

1. Export the existing Modelfile:

ollama show qwen3.6:27b --modelfile > qwen_custom.Modelfile

2. Add or update the num_ctx parameter in the file:

PARAMETER num_ctx 262144

3. Create a new model from it:

ollama create qwen3.6-256k -f qwen_custom.Modelfile

4. Update your client config to point to the new model name, then verify:

ollama ps

The context size shown in ollama ps should now reflect 262144 instead of 32768.

LibreWolf: Claude.ai freezes

If Claude streams responses fine in Firefox but randomly freezes in LibreWolf, you've hit a fingerprinting resistance quirk — not a Claude bug per se:

LibreWolf's Resist Fingerprinting (RFP) deliberately reduces Date.now() precision. This can make consecutive calls return the same value. Claude's frontend assumes time always moves forward — when the delta is zero, it loops forever and the tab freezes.
Create ~/.config/librewolf/librewolf/librewolf.overrides.cfg with the content:
defaultPref("privacy.resistFingerprinting", false);

gentoo: mask the use of new nvidia drivers for legacy cards

Latest nvidia drivers (595.58.03) will not work with old GPUs, e.g. (GTX 960), giving error:
probe with driver nvidia failed
In this case you can mask the newest driver and force gentoo to use older one. You can do this as follows:
echo '>x11-drivers/nvidia-drivers-581' > /etc/portage/package.mask/nvidia-drivers then emerge -avtDUu @world

sudo: how to reset "too many failed attempts" lock

Run as user: faillock --reset

Open WebUI: Run as a temprorary container with Ollama support

docker run --rm --network=host -v open-webui:/app/backend/data -e OLLAMA_BASE_URL=http://127.0.0.1:11434 -e WEBUI_AUTH=False --name open-webui ghcr.io/open-webui/open-webui:main

Distrobox with systemd: "save as" does not work in chromium and brave

If "save as" window does not work in chromium and brave executed in distrobox, this maybe because, container's systemd hijacked debus from the host. Thus you can disable it in the container and restart the container:
systemctl --user mask dbus.socket dbus-broker.service
systemctl --user stop dbus.socket dbus-broker.service
In my case I was using fedora 43 in the container, on gentoo host running systemd and xfce4 (x11). The XDG portal (xdg-desktop-portal-gtk) on host was installed and running.

Firefox: disable "Add appliction" notification for outlook and other email portals

1. Open a new tab.
2. Enter about:config instead of an address.
3. Agree to the warning.
4. In the search bar, enter browser.mailto.dualPrompt and set it to true (double click it to toggle). Next time you see the prompt, it will have a second button called Not now. Press it.


From: https://superuser.com/a/1932569/871038

Qwen-code CLI: Use Qwen 3.6 plus from OpenRouter

I qwen code CLI you can use models from OpenRouter or other providers. For example, to use Qwen 3.6 plus modify your .qwen/settings.json as follows: "env": {
"OPENROUTER_API_KEY": "your-openrouter-api-key"
},
"modelProviders": {
"openai": [
{
"id": "qwen/qwen3.6-plus",
"name": "Qwen3.6-plus (via OpenRouter)",
"envKey": "OPENROUTER_API_KEY",
"baseUrl": "https://openrouter.ai/api/v1",
"generationConfig": {
"timeout": 120000,
"maxRetries": 3,
"samplingParams": {
"temperature": 0.7
}
}
}
]
}

More here.

fedora: web-ext keep changes in a profile for testing and use firefox developer edition

web-ext run -f firefox-aurora --keep-profile-changes --firefox-profile=test

distrobox: firefox "Show in Folder" opens vscodium, not thunar or other file manager.

Run in distrobox (make sure thunar is installed in distrobox first):

xdg-mime default thunar.desktop inode/directory
xdg-mime default thunar.desktop x-scheme-handler/file

bedrock linux: krusader in arch strata does not open files with "Open with" menu from strata

On Bedrock Linux (hijacked gentoo with xfce4) with a non-KDE host session, Krusader's (installed in the arch linux strata) "Open With" silently fails for apps in a guest stratum because KDE's KRun launcher depends on session services that are never started outside a KDE session. The fix is to override the affected .desktop files locally and prepend their Exec lines with strat:
cp /bedrock/strata/arch/usr/share/applications/appname.desktop ~/.local/share/applications/appname.desktop

sed -i 's|^Exec=|Exec=strat arch |' ~/.local/share/applications/appname.desktop

update-desktop-database ~/.local/share/applications/


The simble bash script to do it automatically (fix-strata-desktop.sh) is: #!/bin/bash

STRATA="arch"
STRATA_ROOT="/bedrock/strata/$STRATA"
SRC="$STRATA_ROOT/usr/share/applications"
SRC_CROSS="$STRATA_ROOT/bedrock/cross/applications"
DEST="$HOME/.local/share/applications"

# Icon search dirs within the stratum, in priority order
ICON_SEARCH_DIRS=(
"$STRATA_ROOT/usr/share/icons/hicolor"
"$STRATA_ROOT/usr/share/icons"
"$STRATA_ROOT/usr/share/pixmaps"
)
ICON_EXTENSIONS=("png" "svg" "xpm")
# Preferred sizes in descending order (hicolor subdirs)
ICON_SIZES=("256x256" "128x128" "96x96" "64x64" "48x48" "32x32" "scalable")

# Resolve an icon name to an absolute path inside the stratum.
# Returns the path on stdout, or nothing if not found.
resolve_icon() {
local icon_name="$1"

# Already an absolute path — verify it exists (possibly via stratum root)
if [[ "$icon_name" == /* ]]; then
[[ -f "$icon_name" ]] && echo "$icon_name" && return
[[ -f "$STRATA_ROOT$icon_name" ]] && echo "$STRATA_ROOT$icon_name" && return
return
fi

# Strip any accidental extension for a clean name-based search
local base_name="${icon_name%.*}"

# 1. Search hicolor theme with preferred sizes first
for size in "${ICON_SIZES[@]}"; do
for ext in "${ICON_EXTENSIONS[@]}"; do
local candidate="$STRATA_ROOT/usr/share/icons/hicolor/$size/apps/$base_name.$ext"
[[ -f "$candidate" ]] && echo "$candidate" && return
done
done

# 2. Check pixmaps directly (flat directory, no subdirs) — catches vscodium, cups, etc.
for ext in "${ICON_EXTENSIONS[@]}"; do
local candidate="$STRATA_ROOT/usr/share/pixmaps/$base_name.$ext"
[[ -f "$candidate" ]] && echo "$candidate" && return
done

# 3. Walk all icon theme dirs (any size/category subdir)
for icon_dir in "${ICON_SEARCH_DIRS[@]}"; do
for ext in "${ICON_EXTENSIONS[@]}"; do
# Direct file in dir (pixmaps style — already handled above, but kept for custom dirs)
local candidate="$icon_dir/$base_name.$ext"
[[ -f "$candidate" ]] && echo "$candidate" && return
# Subdir search (themes with size/category layout)
# Prefer largest size by sorting descending on the size component
while IFS= read -r found; do
echo "$found" && return
done < <(find "$icon_dir" -name "$base_name.$ext" 2>/dev/null \
| sort -t'/' -k1,1 -r | head -1)
done
done
}

mkdir -p "$DEST"

for src_file in "$SRC"/*.desktop "$SRC_CROSS"/*.desktop; do
# Skip broken symlinks or non-files
[[ ! -f "$src_file" ]] && echo "Skipping: $(basename "$src_file")" && continue

filename=$(basename "$src_file")
dest_file="$DEST/$filename"

# FIX 1: Remove existing file first so cp never hits a permission-denied
# on files previously written by root (e.g. from an earlier sudo run).
if [[ -e "$dest_file" && ! -w "$dest_file" ]]; then
echo "Warning: removing unwritable existing file: $dest_file"
rm -f "$dest_file" || { echo "Error: cannot remove $dest_file — skipping"; continue; }
fi

cp -- "$src_file" "$dest_file" || { echo "Error: cp failed for $filename"; continue; }

# Fix Exec= lines that don't already use strat
if ! grep -q "Exec=strat $STRATA" "$dest_file"; then
sed -i "s|^Exec=|Exec=strat $STRATA |" "$dest_file"
fi

# Fix Icon= lines — resolve to absolute path inside the stratum
icon_value=$(grep -m1 "^Icon=" "$dest_file" | cut -d'=' -f2-)
if [[ -n "$icon_value" ]]; then
resolved=$(resolve_icon "$icon_value")
if [[ -n "$resolved" ]]; then
# Escape any special characters for sed
escaped=$(printf '%s\n' "$resolved" | sed 's/[[\.*^$()+?{|]/\\&/g')
sed -i "s|^Icon=.*|Icon=$escaped|" "$dest_file"
echo "Processed: $filename (icon -> $resolved)"
else
echo "Processed: $filename (icon '$icon_value' not resolved — left as-is)"
fi
else
echo "Processed: $filename (no Icon= entry)"
fi
done

update-desktop-database "$DEST"
echo "Done."


distrobox: firefox does not open file chooser windows

If you run Firefox inside a Distrobox container on a Gentoo (or any) Linux host, you may notice that the "Save Page As" dialog simply never appears. You hit Ctrl+S, nothing happens. No window, no error, just silence.

The culprit is XDG Desktop Portals. Firefox tries to detect whether it's running inside a sandboxed environment like Flatpak, and when it thinks it is, it hands off file chooser dialogs to the XDG portal service instead of opening a native GTK window. Distrobox leaks enough container-like environment variables to trigger this heuristic, so Firefox dutifully tries to call the portal over D-Bus — but the portal either isn't reachable from inside the container or isn't running at all. The call fails silently and you never see a dialog.

The fix is a single preference change in about:config:
widget.use-xdg-desktop-portal.file-picker = 0
Setting this to 0 tells Firefox to never use the portal for the file picker, falling back to the plain GTK file chooser. The dialog comes back immediately, no restart required.

The default value of 2 means "auto-detect", which is the right call for a real Flatpak sandbox where portals are properly set up. In a Distrobox environment though, the detection fires incorrectly and there's no graceful fallback. Until Distrobox masks the relevant environment variables or Firefox improves its detection logic, flipping this preference manually is the most reliable workaround.

fedora: krusader can't change default application to open files.

Could not find the "keditfiletype" executable in PATH. Install: sudo dnf install keditfiletype

gentoo with systemd: simplescan from flatpak does not open Export save file window

Have to make sure that xdg-desktop-portal-gtk portal on gentoo is installed and running (run a regular user, not root): systemctl --user status xdg-desktop-portal-gtk.service If it is not running enable it: systemctl --user enable --now xdg-desktop-portal-gtk.service

gentoo with systemd and docker: use Nvidia runtime in distrobox through Nvidia container oolkit

Install nvidia container toolbox: sudo emerge -av nvidia-container-toolkit restart docker service: sudo systemctl restart docker Create and setup /etc/cdi: mkdir -p /etc/cdi
nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml Check if its working: nvidia-ctk cdi list Configure docker to use NVIDIA runtime: sudo nvidia-ctk runtime configure --runtime=docker Restart docker service again: sudo systemctl restart docker Before creating distrobox container, lets just perfrom basic check if nvidia is avaiable in docker: docker run --rm --device=nvidia.com/gpu=all ubuntu nvidia-smi If this shows your Nvidia driver and info about your gpu, than it works.

Create distrobox that uses nvidia (fedora 43 as an example): distrobox create -i quay.io/fedora/fedora:43 -n fbox -H ~/home-fbox --additional-flags "--device=nvidia.com/gpu=all"
Enter the distrobox container, fbox, in this example, and test using nvidia-smi: sudo nvidia-smi This should also give info about your nvidia gpu and its driver.

If inside the distrobox container, nvidia is only avaiable for root user, you have to check what group number is nvidia on the host: stat -c "%g" /dev/nvidia0 This will show a number, e.g. 27.
Then in the distrobox container, fbox in this example, you have to edit /etc/group and:
1. add yourself (i.e. normal user) to video group 2. Change video group number to match the one on the host, e.g. 27. For example, /etc/group in fbox
video:x:27:your-user-name

distrobox with nvidia through NVIDIA Container Toolkit

First setup NVIDIA Container Toolkit and docker support for it based on your host's distrubution. Then you can: distrobox create -i archlinux -n abox -H ~/home-abox --additional-flags "--device=nvidia.com/gpu=all" && distrobox enter abox -- bash -c "sudo pacman -Sy --noconfirm git base-devel && git clone https://aur.archlinux.org/yay-bin.git /tmp/yay-bin && cd /tmp/yay-bin && makepkg -si --noconfirm && rm -rf /tmp/yay-bin"

nixos: enable nvidia in distrobox container using Nvidia Container Toolkit

What worked for me on nixos with docker (check nixos wiki on Nvidia and Nvidia Container Toolkit) was:

distrobox create -i archlinux -n archbox --additional-flags "--device=nvidia.com/gpu=all"
Then in the distrobox container you can check if nvidia is detected, using for example glxinfo. It should show: name of display: :0.0
display: :0 screen: 0
direct rendering: Yes
server glx vendor string: NVIDIA Corporation
server glx version string: 1.4
server glx extensions:
Put the following in nixos configuration file: hardware.nvidia-container-toolkit.enable = true;
virtualisation.docker.daemon.settings.features.cdi = true;

Gentoo and SDDM and Ligthdm: Fixing Inactive Mouse at Display Manager Startup in Gentoo with systemd

If you use Gentoo Linux with systemd and a display manager like SDDM or LightDM, you may have run into an annoying issue where the mouse is completely unresponsive on the display manager's login screen after a reboot or cold start. The keyboard works fine, but the mouse does nothing. A second reboot usually fixes it. The problem does not appear on other distributions like NixOS, Fedora, or Ubuntu.

This is a race condition. Gentoo's minimal default configuration does not add unnecessary delays or dependencies to services, which means your display manager can start before the kernel input subsystem and udev have finished initializing the mouse device. Other distributions work around this implicitly through their own service configurations, but on Gentoo you have to handle it yourself.

The fix is to make the display manager wait for udev to finish settling before it starts. Systemd has a built-in service for exactly this purpose called systemd-udev-settle. It blocks until udev has finished processing all queued device events, which includes input devices like your mouse. You wire your display manager to it using a systemd drop-in file.

For SDDM, first create the drop-in directory: mkdir -p /etc/systemd/system/sddm.service.d/

Then create the drop-in configuration file: cat > /etc/systemd/system/sddm.service.d/wait-for-input.conf << 'EOF'
[Unit] After=systemd-udev-settle.service
Wants=systemd-udev-settle.service
EOF

If you use LightDM instead, the steps are identical but with a different directory: mkdir -p /etc/systemd/system/lightdm.service.d/
cat > /etc/systemd/system/lightdm.service.d/wait-for-input.conf << 'EOF'
[Unit] After=systemd-udev-settle.service
Wants=systemd-udev-settle.service
EOF

After creating the file, reload the systemd daemon so it picks up the change: systemctl daemon-reload

That is all. On the next reboot, systemd will start systemd-udev-settle before launching your display manager, and it will wait for all input devices to be fully registered before the login screen appears. The mouse will work on the very first try.

The reason this only affects Gentoo is that you compile and configure everything yourself. There is no distribution maintainer quietly patching service files or adding conservative boot ordering on your behalf. That level of control is what makes Gentoo great, but it also means issues like this surface in a way they simply would not on a more opinionated distribution.

If the problem persists after applying this fix, the next thing to investigate is whether your mouse and evdev drivers are compiled as kernel modules rather than built into the kernel, and whether INPUT_DEVICES is correctly set in your /etc/portage/make.conf.

Fedora 43: Install latest foxitreader

Download latest version (FoxitReader.enu.setup.2.4.5.0727). Foxitreader is no longer maintained, so this is the oldest version.
wget https://cdn01.foxitsoftware.com/pub/foxit/reader/desktop/linux/2.x/2.4/en_us/FoxitReader.enu.setup.2.4.5.0727.x64.run.tar.gz
tar xzvf FoxitReader*.tar.gz
chmod +x ./"FoxitReader.enu.setup.2.4.5.0727(rb70e8df).x64.run"
sudo ./"FoxitReader.enu.setup.2.4.5.0727(rb70e8df).x64.run"
Follow the prompts.

gentoo with xfce4: krusader does not start from distrobox

kf.i18n: KLocalizedString: Domain is not set for this string, translation will not work. Please see https://api.kde.org/frameworks/ki18n/html/prg_guide.html msgid: "No jobs" msgid_plural: "" msgctxt: ""

Authorization required, but no authorization protocol specified

10:26:41.584-warning qt.qpa.xcb unknown@0 # could not connect to display :0.0 10:26:41.584-warning qt.qpa.plugin unknown@0 # From 6.5.0, xcb-cursor0 or libxcb-cursor0 is needed to load the Qt xcb platform plugin. 10:26:41.584-info qt.qpa.plugin unknown@0 # Could not load the Qt platform plugin "xcb" in "" even though it was found. 10:26:41.584-fatal default unknown@0 # This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: linuxfb, wayland-brcm, wayland-egl, wayland, vkkhrdisplay, vnc, offscreen, xcb, minimalegl, eglfs, minimal.
install xhost in gentoo host: sudo emerge x11-apps/xhost
and allow access to display: xhost +
now can enter the fedora distrobox

bedrock with hijacked gentoo and xfce4: missing icons from strata in xfce4 application menu

sudo touch /usr/share/icons/hicolor
sudo gtk-update-icon-cache
based on: https://old.reddit.com/r/bedrocklinux/comments/llt5pp/buggy_icons_from_other_stratums_in_gnome/gnuo7tv/

gentoo and libvirt: Unable to find a satisfying virtiofsd

emerge --ask app-emulation/virtiofsd

tigervnc with fluxbox example setup

/etc/tigervnc/vncserver-config-mandatory session=startfluxbox
securitytypes=none
geometry=2000x1200
localhost
alwaysshared


Gentoo install app-arch/ rar - license

To avoid license inssues in future, you can just accept all licenses in /etc/portage/make.conf: ACCEPT_LICENSE="*"

gentoo: list recently updated packages

Emerge genlop if needed, and then: sudo genlop -u

distrobox: create archlinux container with automatic yay installation

distrobox create -i archlinux -n abox -H ~/home-abox && distrobox enter abox -- bash -c "sudo pacman -Sy --noconfirm git base-devel && git clone https://aur.archlinux.org/yay-bin.git /tmp/yay-bin && cd /tmp/yay-bin && makepkg -si --noconfirm && rm -rf /tmp/yay-bin"

distrobox: start krusader with dark them from host

/run/current-system/sw/bin/distrobox-enter -n abox -- bash -c 'env GTK_THEME=Adwaita:dark ICON_THEME=Papirus QT_QPA_PLATFORMTHEME=qt6ct krusader -qwindowtitle %c %u'

Fixing KDE App Theming in Distrobox on XFCE

If you run a KDE application (like Krusader) inside a Distrobox container on an XFCE host with a dark theme, you might notice that the icons are dark and nearly invisible against the dark background. This happens because the app inherits a mismatched icon theme from the host, and no Qt platform theme manager is configured inside the container.

The Fix

First, enter your container and install the required packages:

distrobox enter abox
sudo pacman -S papirus-icon-theme qt6ct


Note: check whether your KDE app links against Qt5 or Qt6 before choosing between qt5ct and qt6ct:

ldd $(which krusader) | grep -i qt


Next, tell Qt to use qt6ct as the platform theme. In fish shell, set it as a universal variable so it persists across sessions:

set -Ux QT_QPA_PLATFORMTHEME qt6ct


Then configure the theme and icons by editing the qt6ct config directly:

mkdir -p ~/.config/qt6ct
cat > ~/.config/qt6ct/qt6ct.conf << 'EOF'
[Appearance]
icon_theme=Papirus-Dark
style=Breeze
EOF


Finally, set up kdeglobals so KDE apps pick up the correct color scheme and icon theme:

cat > ~/.config/kdeglobals << 'EOF'
[General]
ColorScheme=BreezeDark

[KDE]
widgetStyle=Breeze

[Icons]
Theme=Papirus-Dark
EOF


Open a fresh container session and launch your KDE app — the icons and theme should now render correctly.

terminator: change its window size

terminator --geometry 2400x650+100+1200

volid linux: install hplip and enable cupsd

sudo xbps-install -S hplip sudo ln -s /etc/sv/cupsd /var/service sudo sv status cupsd

shell.nix for golang programming with vscodium

If you use Nix and want a reproducible Go development environment with VSCodium as your editor, a shell.nix file is the cleanest way to get there. The configuration below sets up a full Go toolchain — including the gopls language server, the delve debugger, linters, and code-generation utilities — alongside VSCodium pre-loaded with the official Go extension, GitLens, error highlighting, and a REST client. CGO is enabled together with gcc and sqlite, so packages like go-sqlite3 that rely on C bindings compile without issues. A step-by-step status log is printed during initialisation so you always know what the shell is doing while it loads. All dependencies are pinned to your nixpkgs channel, so every team member or machine gets an identical environment simply by running nix-shell. Once inside the shell, open your project with codium . and everything is ready to go.

{ pkgs ? import <nixpkgs> {
# Required to allow extensions with non-free licenses (e.g. gitlens)
config.allowUnfree = true;
}
}:

let
vscodiumWithExtensions = pkgs.vscode-with-extensions.override {
vscode = pkgs.vscodium;
vscodeExtensions = with pkgs.vscode-extensions; [
# Go language support (official Go team extension)
golang.go

# Git integration
# Note: mhutchie.git-graph is unfree and has been intentionally removed.
# Use VSCodium's built-in git view or install git-graph manually.
eamodio.gitlens

# General productivity
streetsidesoftware.code-spell-checker
usernamehw.errorlens

# REST client for API testing
humao.rest-client
];
};

in pkgs.mkShell {
name = "golang-dev";

buildInputs = [
# ── Go toolchain ──────────────────────────────────────────────────────────
pkgs.go # Go compiler & standard tooling (go build, go test…)
pkgs.gopls # Official Go language server (used by the extension)
pkgs.delve # Go debugger
pkgs.gotools # Extra tools: goimports, godoc, etc.
pkgs.golangci-lint # Fast, multi-linter runner
pkgs.gomodifytags # Struct tag editor (used by vscode-go)
pkgs.gotests # Auto-generate table-driven tests
pkgs.impl # Generate method stubs for interfaces
pkgs.go-outline # JSON representation of Go file outline
pkgs.gotestsum # Prettier `go test` output

# ── Editor ────────────────────────────────────────────────────────────────
vscodiumWithExtensions

# ── C toolchain (required for CGO / go-sqlite3) ───────────────────────────
pkgs.gcc
pkgs.sqlite

# ── Shell / system utilities ──────────────────────────────────────────────
pkgs.git
pkgs.curl
pkgs.jq # Handy for JSON output while developing APIs
];

# ── Environment variables ─────────────────────────────────────────────────
shellHook = ''
echo ""
echo "● Initialising golang-dev shell…"
echo ""

# ── Environment ───────────────────────────────────────────────────────────
echo " [1/4] Setting up environment variables…"
export GOPATH="$HOME/go"
export PATH="$GOPATH/bin:$PATH"
export GOTELEMETRY=off
export CGO_ENABLED=1
echo " GOPATH=$GOPATH"
echo " CGO_ENABLED=$CGO_ENABLED (sqlite3 / cgo support)"
echo " GOTELEMETRY=$GOTELEMETRY"

# ── Go toolchain ──────────────────────────────────────────────────────────
echo ""
echo " [2/4] Verifying Go toolchain…"
echo " go $(go version 2>/dev/null | awk '{print $3, $4}')"
echo " gopls $(gopls version 2>/dev/null | head -1 || echo 'not found')"
echo " delve $(dlv version 2>/dev/null | head -1 || echo 'not found')"
echo " golangci $(golangci-lint --version 2>/dev/null | head -1 || echo 'not found')"
echo " gcc $(gcc --version 2>/dev/null | head -1 || echo 'not found')"
echo " sqlite3 $(sqlite3 --version 2>/dev/null | head -1 || echo 'not found')"

# ── GOPATH bin ────────────────────────────────────────────────────────────
echo ""
echo " [3/4] Checking GOPATH bin directory…"
if [ -d "$GOPATH/bin" ]; then
BIN_COUNT=$(ls "$GOPATH/bin" 2>/dev/null | wc -l | tr -d ' ')
echo " $GOPATH/bin ($BIN_COUNT tools installed)"
else
mkdir -p "$GOPATH/bin"
echo " $GOPATH/bin created (empty — tools will appear after go install)"
fi

# ── Editor ────────────────────────────────────────────────────────────────
echo ""
echo " [4/4] Checking VSCodium…"
if command -v codium &>/dev/null; then
echo " $(codium --version 2>/dev/null | head -1 | sed 's/^/codium /')"
else
echo " codium not found in PATH"
fi

# ── Ready ─────────────────────────────────────────────────────────────────
echo ""
echo "┌─────────────────────────────────────────┐"
echo "│ golang-dev shell ready ✓ │"
echo "│ │"
echo "│ Run 'codium .' to open VSCodium │"
echo "└─────────────────────────────────────────┘"
echo ""
'';
}


Gwenview: can't open TIFF images

sudo pacman -S qt6-imageformats kimageformats

distrobox: export Notepad++ installed thorugh wine

distrobox-export --app Notepad++

void linux with docker: distrobox hangs when a container with --init and systemd is created.

Distrobox lets you run any Linux distribution as a container and integrate it into your host system. But if your host is Void Linux, getting a systemd-based box like Arch to boot correctly takes a bit of extra work.

The problem

Void uses runit as its init system, not systemd. When you create a distrobox with --init and ask for a systemd container, it will hang forever printing:

waiting for systemd to come up...
waiting for systemd to come up...
waiting for systemd to come up...

Even after fixing the cgroup setup (which also requires attention on a non-systemd host), systemd inside the container will fail on several units that try to mount kernel filesystems — things like binfmt_misc, hugepages, and sys-kernel-debug. These units make no sense inside a container and will block the boot sequence.

The fix

The solution is to mask those units at creation time, before systemd ever tries to start them. The --init-hooks flag runs a shell command during container setup, so we can pre-create the /dev/null symlinks that systemctl mask would normally create:

distrobox create -i archlinux -n abox -H ~/home-abox --init --additional-packages "systemd" --additional-flags "--privileged" --init-hooks "mkdir -p /etc/systemd/system && for u in proc-sys-fs-binfmt_misc.automount proc-sys-fs-binfmt_misc.mount sys-kernel-debug.mount sys-kernel-tracing.mount dev-hugepages.mount systemd-firstboot.service systemd-udevd.service; do ln -sf /dev/null /etc/systemd/system/\$u; done"

The --privileged flag gives systemd the capabilities it needs to manage its own cgroup namespace. The hook symlinks all the container-hostile units to /dev/null, which is exactly what masking does — just without needing systemctl to be running yet.

Why symlinks instead of systemctl? At hook execution time systemd hasn't started yet, so systemctl mask may not work reliably. Writing symlinks directly achieves the same result and works unconditionally.

Result

After running this command, distrobox enter abox will boot straight through into a fully working Arch Linux shell with a live systemd instance — on a Void Linux host running runit.

Posted March 2026

Void Linux: "/" is not a shared mount, this could cause issues or missing mounts with rootless containers

When running distrobox with podman:
"/" is not a shared mount, this could cause issues or missing mounts with rootless containers Just add mount --make-rshared / to your /etc/rc.local

Void Linux, podman and distrobox: unable to apply cgroup configuration

Error: unable to start container "0a58490ca8d4b9cf464483b8a776e53616ebfc60cc89aa25e60b0d910d907eb2": runc: runc create failed: unable to start container process: unable to apply cgroup configuration: rootless needs no limits + no cgrouppath when no permission is granted for cgroups: mkdir /sys/fs/cgroup/0a58490ca8d4b9cf464483b8a776e53616ebfc60cc89aa25e60b0d910d907eb2: permission denied: OCI permission denied
Create ~/.config/containers/containers.conf: [containers]
cgroupns = "host"


[engine]
cgroup_manager = "cgroupfs"
runtime = "crun"

sudo xbps-install -S crun

Remove the container, and install it again, e.g.: distrobox stop abox
distrobox rm abox
podman system reset
distrobox create -i archlinux -n abox -H ~/home-abox
distrobox-enter abox


Void Linux: podman and distrobox: short-name "archlinux" did not resolve to an alias and no unqualified-search registries are defined in "/etc/containers/registries.con

Error: short-name "archlinux" did not resolve to an alias and no unqualified-search registries are defined in "/etc/containers/registries.conf"
Edit: /etc/containers/registries.conf and add:

unqualified-search-registries = ["docker.io"]

Fix: Firefox in Distrobox "Show in Folder" Opens the Wrong App on Linux

In Distrobox, install thunar if using xfce4 and add inode/directory=thunar.desktop into ~/.config/mimeapps.list: [Default Applications]
inode/directory=thunar.desktop

Start libreoffice in dark theme from distrobox on nix os through .desktop file

On Host, after exporting the libreoffice from distrobox, edit e.g. ~/.local/share/applications/abox-libreoffice-calc.desktop and add "env GTK_THEME=Adwaita:dark": Exec=env GTK_THEME=Adwaita:dark /run/current-system/sw/bin/distrobox-enter -n abox -- libreoffice --writer %U

flatpak: Kdenlive error: Could not initialize GLX

Could not initialize GLX Have to **update flatpak** as probably nvidia drivers in flatpaks are much different than on the host.

NixOS: Libreoffice in distrobox does not use Adwaita dark theme

GTK_THEME=Adwaita:dark libreoffic
or set in fish config:
set -x GTK_THEME Adwaita:dark

NixOS: Firefox in distrobox does not use Adwiata theme (no hand mouse cursor)

NixOS injects its Nix store paths into XCURSOR_PATH so the host system finds cursors in /nix/store/..., and distrobox inherits that variable. Inside the Arch container those paths don't exist, so libxcursor found nothing and silently fell back to the default arrow for every cursor change.
vim .config/fish/config.fish
set -e XCURSOR_PATH
set -x XCURSOR_PATH /usr/share/icons ~/.icons


When creating launcher for firefox in xfce4, have to reset `XCURSOR_PATH`:
/run/current-system/sw/bin/distrobox-enter -n abox -- bash -c 'XCURSOR_PATH=/usr/share/icons /usr/lib/firefox/firefox %u '

terminator: change size and location

terminator --geometry=2420x650+80+800

NixOs: Can't start terminator from distrobox on nixos

TypeError: unknown signal name: changed::font-name

The error TypeError: unknown signal name: changed::font-name occurs because Terminator is trying to connect to a GSettings signal for org.gnome.desktop.interface (font settings), but the required GSettings schemas are missing or inaccessible inside the distrobox container.
distrobox-enter -n abox -- bash -c 'unset LD_LIBRARY_PATH GI_TYPELIB_PATH GIO_MODULE_DIR GSETTINGS_SCHEMA_DIR; terminator'

distrobox-assembly: Install arch with yay and some default packages

distrobox.ini:
[abox]
additional_packages="git base-devel sudo curl"
home=~/home-abox
image=archlinux:latest
init=true
start_now=true
init_hooks='useradd -m -G wheel builder && echo "%wheel ALL=NOPASSWD: ALL" >> /etc/sudoers.d/wheel && su - builder -c "git clone https://aur.archlinux.org/yay-bin.git && cd yay-bin && makepkg -si --noconfirm" && rm -rf /home/builder/yay-bin && su - builder -c "yay -S --noconfirm --needed mc vim htop tmux fd zoxide fzf terminator freerdp rdesktop bash-completion firefox firefox-developer-edition brave-bin vscodium-bin onlyoffice-bin libreoffice-fresh gwenview okular wine krusader krename kdiff3 konsole xdg-desktop-portal-kde ipython tigervnc fluxbox rofi xterm xorg-xeyes masterpdfeditor vlc mpv qt6-imageformats kimageformats vlc-plugins-all fish chromium librewolf-bin go python-uv gimp ttf-roboto otf-font-awesome rar p7zip octave yt-dl sshfs ebook-tools xarchiver kdegraphics-mobipocket gnome-boxes" && cd /etc/xdg/menus && ln -sf ./plasma-applications.menu applications.menu && ln -sf ./plasma-applications.menu gnome-applications.menu && ln -sf ./plasma-applications.menu kde-applications.menu && su - builder -c "NONINTERACTIVE=1 /bin/bash -c \"\$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)\" && eval \"\$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)\" && brew install claude-code gemini-cli opencode qwen-code web-ext" && echo "eval \"\$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)\"" >> ~/.bashrc
nvidia=true
pull=true
root=false
replace=false


Fedora atomic (vauxite xfce4): install nvidia drivers

sudo rpm-ostree install akmod-nvidia xorg-x11-drv-nvidia
sudo rpm-ostree kargs --append=rd.driver.blacklist=nouveau,nova_core --append=modprobe.blacklist=nouveau,nova_core --append=nvidia-drm.modeset=1

https://github.com/eoladil/Fedora-Silverblue-Guides?tab=readme-ov-file#4-install-nvidia-drivers-for-current-geforce-tesla-and-quadro-gpus

tigvervnc: example vncserver config setup

vncserver-config-mandatory:

session=startfluxbox
securitytypes=none
geometry=1800x1100
localhost
alwaysshared

kde plasma: set kde theme from a command line in a distrobox

List current themes:

lookandfeeltool -list
can return

org.kde.breeze.desktop
org.kde.breezedark.desktop
org.kde.breezetwilight.desktop
then

lookandfeeltool -a org.kde.breezedark.desktop

More at: https://askubuntu.com/a/1183309

Bazzite (Fedora atomic): install virtualbox on Bazzite custom image

Bazzite has rpm fusion repositories added in /etc/yum.repos.d/. However, even though they are enabled, by default you can't install any packages from them. You can force the use of the RPM repo as follows:

sudo dnf --enablerepo=rpmfusion-free install virtualbox
This will still NOT install the virtualbox in atomic fedora as its read-only system. But can be useful for building custom atomic images which would contain the virtualbox or other packages.

Example of how the virtualbox can be installed in a custom image of Bazzite is here:
https://github.com/ettfemnio/bazzite-virtualbox

distrobox: Basic programs for archlinux and fedora in distrobox

yay --noconfirm -S mc vim fish htop tmux fd zoxide fzf terminator freerdp rdesktop bash-completion firefox firefox-developer-edition brave-bin vscodium-bin libreoffice-fresh gwenview okular wine krusader krename kdiff3 konsole sshfs xarchiver xdg-desktop-portal-kde tigervnc fluxbox rofi xterm xorg-xeyes vlc mpv vlc-plugins-all chromium ipython librewolf-bin go python-uv gimp ttf-roboto otf-font-awesome rar p7zip octave ebook-tools kdegraphics-mobipocket gnome-boxes qt6-imageformats kimageformats curl wget foxitreader thunar npm kate

sudo dnf -y install --skip-unavailable vim firefox brave-browser krusader krename konsole kate keditfiletype kdiff3 chromium okular vlc htop tmux mc fish go python-uv codium unrar p7zip nvtop glx-utils xfreerdp rdesktop tigervnc libreoffice gwenview wine sshfs xarchiver xdg-desktop-portal-kde mpv gnome-boxes curl wget fzf zoxide terminator fontawesome-fonts-all librewolf gimp thunar octave yazi nodejs kwrite firefox-devedition Note: Will have to add custom repos to fedora for firefox-dev, brave-browser, codium and rpmfusion for codeks

Use GLM through OpenRouter in Claude Code

ANTHROPIC_AUTH_TOKEN="your-openrouter-api-key" ANTHROPIC_BASE_URL="https://openrouter.ai/api" ANTHROPIC_API_KEY="" claude --model z-ai/glm-4.7

arch: Uninstall nvidia-open drivers due to second monitor being not detected

sudo pacman -Rd nvidia-open nvidia-utils

Systemd: autostart Ansys SAM

sudo vim /etc/systemd/system/sam.service


[Unit]
Description=Ansys SAM Web Application Container
Requires=docker.service
After=docker.service

[Service]
Type=simple
User=root
Group=root
WorkingDirectory=/opt/ansys-mbse-server/mbse
ExecStartPre=/usr/bin/docker compose down
ExecStart=/usr/bin/docker compose up
ExecStop=/usr/bin/docker compose down
Restart=always
TimeoutStopSec=30

[Install]
WantedBy=multi-user.target


SSL certificate keystore for Ansys SAM tomcat server

Getting SSL certificate for a domain using certbot (https://certbot.eff.org/) gives you four files: cert1.pem, chain1.pem, fullchain1.pem, privkey1.pem. These files can found in /etc/letsencrypt/archive/your.domain.name.

The files have to be bundled into a keystore (keystore.pfx) for use in tomcat. For this you can use:

openssl pkcs12 -export -in cert1.pem -inkey privkey1.pem -out keystore.pfx -name tomcat -CAfile chain1.pem -caname root -chain
"keystore.pfx" must be renamed to "keystore" and keystore password used when creating keystore.pfx have to be specified in server.xml.

More information at: https://hardwarehacks.org/blog/setting_up_tomcat_with_a_lets_encrypt_certificate/

ssh: LocalForward in ssh config file

Example to forward vnc 5901 port from server to client's 5903 port:
Host myserver
    HostName serverip
    User ubuntu
    IdentityFile ~/.ssh/private.pem
    LocalForward 5903 localhost:5901


tigervnc: on ubuntu example settings

/etc/tigervnc/vncserver-config-mandatory :
$session = "startfluxbox";
$SecurityTypes = "none";
$geometry = "2000x1200";
$localhost = "yes";
$AlwaysShared = "yes";


sudo systemctl enable --now tigervncserver@:1