C-u usually erases everything from the start of line to the cursor.

$ magick -delay 10 -loop 0 -dispose Background *.png output.gif

Delay is in ms.

After the expiration of my emails.myserver1.com and emails.myserver2.com certificates, I had troubles with some mail client to send mails. As a temporary solution I have configured msmtp (~/.config/msmtp/config) disabling certificates verification but keeping fingerprint checking. Here is how to check for fingerprints:

for server in "myserver1" "myserver2"; do
    openssl s_client -connect emails.${server}.com:465 < /dev/null 2>/dev/null | openssl x509 -fingerprint -noout -in /dev/stdin
done

And in ~/.config/msmtp/config I have put tls_fingerprint <fingerprint>.

In /etc/cron.d/certbot I added 0 */12 * * * root certbot renew --no-random-sleep-on-renew && systemctl restart postfix dovecot

opkg list-upgradable | cut -f 1 -d ' ' | xargs -r opkg upgrade  

Some links cannot be seen with pdftotext. I used this trick from SO:

$ nb_pages=220 # set the number of pages of pdf
$ pdftk file.pdf cat 1-$nb_pages output - | strings | grep -E "https?://"

$ USERNAME=bouzin
$ DATABASE_NAME=something
$ mysqldump -u $USERNAME -p $DATABASE_NAME > dump.sql

sort -uo file.txt{,}

echo "hello world" | tee >(grep -o world >> output.txt)

This uses process substitution, and will print “hello world” on the screen, but keep only “world” in output.txt.

# display web history sorted by date
$ sqlite3 ~/.mozilla/firefox/xxxx.default-release-xxxx/places.sqlite "SELECT datetime(moz_historyvisits.visit_date/1000000,'unixepoch'), moz_places.url FROM moz_places, moz_historyvisits WHERE moz_places.id = moz_historyvisits.place_id"|sort
 
# display all unique visited urls
$ sqlite3 ~/.mozilla/firefox/xxxx.default-release-xxxx/places.sqlite "SELECT url FROM moz_places"

https://aurelieherbelot.net/web/read-firefox-history-linux-terminal/

Something I have used to modify firmwares when I was tweaking the NB6VAC. Use xxd in vim:

:%! xxd -b # this convert the file to binary. make your changes then
:%! xxd -r # this reverts

I have used a man in the middle technique to try to capture traffic from my Android phone. (Goal: intercept packets from a CloudEdge application I have now good reason not trusting…)

1. Start kali linux, be sure every devices are on same WiFi (!)
2. sudo airmon-ng start wlan1
3. wireshark > capture on “wlan0” interface
4. filter: “ip.addr==my_phone_ip”
5. sudo ettercap
6. “scan for hosts”, “hosts list” > select router IP (target1) and phone IP (target2)
7. “current targets”
8. “MITM” > “start ARP poisoning”
9. wireshark should start showing traffic

• https://openmaniak.com/ettercap_arp.php
• https://stackoverflow.com/questions/9555403/capturing-mobile-phone-traffic-on-wireshark

Interactive comments is by default on Bash. It allows using comments when writing commands in the terminal.

# ~/.config/zsh/.zshrc
setopt interactivecomments

Now you can write echo "this is printed" # this is not

Add this to the YAML header options:

header-includes:
- |
    ```{=latex}
    \usepackage{fvextra}
    \DefineVerbatimEnvironment{Highlighting}{Verbatim}{breaklines,breakanywhere,commandchars=\\\{\}}
    ```

https://tex.stackexchange.com/a/412316

$ git add --patch file/to_patch

$ git add --help
...
   patch
       This lets you choose one path out of a status like selection. After choosing the path, it presents the diff between the index and the working tree file and asks you if you want to stage the change of each hunk. You can select one of the following options and type return:
 
           y - stage this hunk
           n - do not stage this hunk
           q - quit; do not stage this hunk nor any of the remaining ones
           a - stage this hunk and all later hunks in the file
           d - do not stage this hunk nor any of the later hunks in the file
           g - select a hunk to go to
           / - search for a hunk matching the given regex
           j - leave this hunk undecided, see next undecided hunk
           J - leave this hunk undecided, see next hunk
           k - leave this hunk undecided, see previous undecided hunk
           K - leave this hunk undecided, see previous hunk
           s - split the current hunk into smaller hunks
           e - manually edit the current hunk
           ? - print help

User-agent: *
Disallow: /

https://www.robotstxt.org/robotstxt.html

After copying the “new” mkinitcpio.conf.pacnew onto my “old” and running mkinitcpio -P blindly, I got this error on restart, in the uncommon “emergency shell”:

/dev/mapper/vgarchlinux-root not found, skipping fsck

Too bad. I downloaded an arch.iso, ran my dd command and ran this to solve:

$ mount /dev/mapper/vgarchlinux-root /mnt
$ cd /mnt
$ mount -t proc /proc proc/ # necessary to run mkinitcpio in chroot
$ mount --rbind /sys sys/   # necessary to run mkinitcpio in chroot
$ mount --rbind /dev dev/   # necessary to run mkinitcpio in chroot
$ chroot .
$$ mkinitcpio -P

And voilà!

$ gpasswd --delete user group

First be sure to remove this old symbolic link:

$ sudo rm /usr/lib/surf/libsurf-webext.so

Then either:

$ WEBKIT_DISABLE_DMABUF_RENDERER=1 && surf

from source

Or better:

$ cat /sys/module/nvidia_drm/parameters/modeset
N # <- if N we need to enable
$ echo options nvidia_drm modeset=1 | sudo tee /etc/modprobe.d/nvidia_drm.conf

• png: optipng image.png
• jpeg: jpegoptim image.jpeg

$ start_at_line=100 # whatever
$ hashcat "hashex.txt" "wordlist.txt" -s ${start_at_line}

$ grep -aP "[\x80-\xFF]" file.txt

source

$ vim example.txt
...
abc^M
defg^M
...

source

To remove this, enter this in vim:

:e ++ff=dos
:set ff=unix
:wq

This forces vim to re-read the file forcing dos file format. Then we set it back to unix file format.

rlwrap -A -pgreen -S"perl> " perl -wnE'say eval()//$@'

I had a problem when fetching a git repo I made on my LFS system: it had stored my real email address and I could not push it to GitHub (it is GitHub that warned me).

To solve this I had to set my “no-reply” address in the local repo:

git config user.email "blabla@users.noreply.github.com"
git commit --amend --reset-author --no-edit
git push

This is my current goto for modules in Perl.

Let's say I have main.pl and my_module.pl, where the module contains only functions.

We need to add 1; at the end of the module so it is evaluated to true when imported into main.pl.

Then, in main.pl, we import with this line : require "./my_module.pl".

It is important to know that functions called from my_module into main need parenthesis for their argument (even when they do not take any!).

For example: a_function_from_module();.

:find filename

:b filename # (can be truncated)

git log -p --full-diff -- filename

In FreeCAD:

• select every pieces of the object we want (Shift+left_click)
• select the Part menu
• then Part > Compound > Create compound

Now the object is a single object and can easily be imported in Prusa.

$ cat file.txt
1.2
12.4
0.2
...
$ gnuplot
$ plot "file.txt" with lines

$ sbcl --load hello.lisp
* (load "hello.lisp")
* (require "asdf")
* (setq uiop:*image-entry-point* #'main) ; imagining that entry point is function "main"
* (uiop:dump-image "hello.exe" :executable t)

• https://stackoverflow.com/questions/14171849/compiling-common-lisp-to-an-executable

Another (own) way to compile a lisp file:

$ sbcl --eval '(compile-file "my_script.lisp")' --eval '(quit)'
$ chmod +x ./my_script.fasl
$ ./my_script.fasl

Inspired from here:

$ sbcl
* (compile-file "my_program.lisp")
* (load "my_program.fasl")

Else, directly in a REPL: (load (compile-file "my_program.lisp")).

I had trouble with an update (package xp-pen-tablet). It broke my pen tablet.

To revert back, I did :

git clone https://aur.archlinux.org/xp-pen-tablet.git
cd ./xp-pen-tablet.git
git revert --no-commit eef350b9b9bbdbc43ebd18e19f49b99869977d53 HEAD
git commit
makepkg -i

:runtime syntax/2html.vim

https://dud.univ-amu.fr/fr/public/quelques-killer-tips-vim

netstat -n --program | grep "firefox"

zcat /proc/config.gz

https://unix.stackexchange.com/questions/176307/where-can-i-find-the-kernel-configuration-on-each-linux-distribution

#!/bin/sh
username="$1"
curl -sL "https://twitch.tv/$username" | grep -o "isLiveBroadcast" && notify-send "$username is live!"

export MENUCONFIG_COLOR=mono

But… a better (?) alternative for me would be to use : make nconfig ! It has vim keybindings!

https://github.com/crosstool-ng/crosstool-ng/issues/1175

I tend to use grep and sed together : to match, then to replace. But we can match with sed :

echo "Hello world" | sed "/w/ s/world/planet/" file.txt
Hello planet

echo "Hello world" | sed "/z/ s/world/planet/" file.txt
Hello world

This allow to replace only when a line matched /w/. Note that it doesn't replace when trying to match “Hello world” with /z/.

sed -i.bak "s/this/that/" file.txt

This will write the changes inside file.txt, but also create a backup of it in file.txt.bak.

Note that this work for multiple files too :

sed -i.bak "s/this/that/" file1.txt file2.txt file3.txt

To prefix the backup “extension” : -i'bak.* will produce bak.file1.txt, bak.file2.txt, etc.

https://learnbyexample.github.io/learn_gnused/in-place-file-editing.html

C-x-; = comment/uncomment lines. C-x-x = reselect previous selection

After git merge source destination some files might be in conflict. To solve this, modify conflicting files, then git add the_files and git rebase --continue.

Squashing allows to compress multiple previous commits together. git rebase -i HEAD~n note that -i stands for interactive, and n should be an integer corresponding to the number of commits that should be merged together. To know n, check git log and decide for yourself. After squashing commits (squash every commit under one), git push --force origin will send it to your working distant branch.

That is especially useful when you made changes to a branch you did not want to change. git checkout -b newbranch will create and send non-commited changes to newbranch, and (!) will let untouched the current one!

previous_commits_to_squash=3 #check with git log to know how much should be squashed
git rebase -i HEAD~"${previous_commits_to_squash}"
# rename to "squash" for every commit under the first commit
# merge comments
git push --force origin

find <path> -daystart -ctime 0 -print

For 1 hour:

find <path> -cmin -60 # change time
find <path> -mmin -60 # modification time
find <path> -amin -60 # access time

echo apple1_blabbla | sed '/apple./ { s/apple1/apple2/g; }'
echo apple1 | sed -e 's/\(a\)\(p*\)\(le\)1/\1\2\32/g' # use capture group!!

Add a user to existing file :

sudo htpasswd /etc/apache2/passwd/<FILE> <username>

else, Create file :

sudo htpasswd -c /etc/apache2/passwd/<FILE> <username>

then in /etc/apache2/sites-enabled/XXX.conf append this to the website:

<Location />
    AuthType Basic
    AuthName "Text to greet with :)"
    AuthBasicProvider file
    AuthUserFile "/etc/apache2/passwd/passwords"
    Require user john, patrick
</Location>

ffmpeg -i input.mkv -vcodec libx265 -crf 28 output.mp4

See man 5 alpm-hooks.

I had the following persistent problem : when rebuilding Linux initcpios (rebuilding initramfs images), my keyboard bindings (keymaps) where reset to default. This resulted in me not having the correct Super Key anymore. A way to solve that was creating a pacman hook in /etc/pacman.d/hooks/reload-keyboard-mappings.hook with:

[Trigger]
Type=Path
Operation=Upgrade
Operation=Install
Target = boot/initramfs-linux*
Target = usr/lib/modules/*/vmlinuz
Target = usr/lib/initcpio/*
Target = usr/lib/firmware/*
Target = usr/src/*/dkms.conf
 
[Trigger]
Type=Package
Operation=Upgrade
Operation=Install
Target = linux
Target = linux-*
Target = nvidia*
 
[Action]
Description=Setting keyboard layout after updating initramfs
When=PostTransaction
Exec=/home/me/.local/bin/remaps pacman_hook

csplit -z thisfile.md /^#\ / '{*}' # pattern = line starting with # followed by a space
#csplit -z <FILE> <pattern> '{*}'

-z removes empty output files.

paru -S magic-wormhole
wormhole send FILE # on the sender
wormhole receive   # on the receiver

Following the ArchWiki install guide, I saw that line : ls /usr/share/kbd/keymaps/**/*.map.gz. What does the ** mean? It means matches filenames and directories recursively. See here for small explaination.

Thus, if there are files like /usr/share/kbd/keymaps/amiga/xxxx.map.gz and /usr/share/kbd/keymaps/mac/all/xxxx.map.gz, the command ls /usr/share/kbd/keymaps/**/*.map.gz will find those. Even if directories' trees are different.

See here

By default, Terminal starts the shell via /usr/bin/login, which makes the shell a login shell. On every platform (not just Mac OS X) bash does not use .bashrc for login shells (only /etc/profile and the first of .bash_profile, .bash_login, .profile that exists and is readable). This is why "put source ~/.bashrc in your .bash_profile" is standard advice

for c in $(docker ps -q); do
    docker inspect $c -f "{{ .Name }}:";
    docker top $c | awk '{print $1, $2, $8}';
    echo "--------------";
done

However I doubt how I should check for security concerns. Lots of website state that “docker should not be run as root” but they mean the process INSIDE the container as the docker daemon dockerd is forced to be run as root. The problem for me is that it seems that some container need their init process to be run as root. So how do I know?

Check the command netstat (with the command watch).

More precisely use to extract only connected IPs netstat -rntu | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -n.

Some website recommand netstat -tulpn.

To watch it in real time : watch netstat -tulpn. That's where the watch program shows its interest! For commands with awk, we need to put the command in quotes: watch "netstat -rntu | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -n".

https://bbs.archlinux.org/viewtopic.php?id=261128

Yann : “Got it. I took a time to study the X clipboard fonctionnement. I did not expect that. xclip or xsel are X processes without X windows. When using them to set the clipboard, they are just X windows owning the clipboard in their own buffer. When another X application needs to past, the X server will call a specific function of the X application owning the clipboard to send the application buffer content containing the copy to the stdin of the pasting X application. Why not having this buffer in the X server directly? This forces every X application to have the necessary functions to support this feature. Also, when closing the X application owning the clipboard, it is gone. I could see the -loop X argument in action, when X pasts are done, the xclip process gets killed and there is no more clipboard available.”

Trilby : “Yann wrote:

This forces every X application to have the necessary functions to support this feature.

False. Only programs that intend to interact “copy” or “cut” data to a selction need these functions, and the functions are in fact quite simple; or at least they can be, some programs may have elaborate versions, but the basic requirements are trivial. Yann wrote:

Also, when closing the X application owning the clipboard, it is gone.

This is true. And this is one of the primary motivations of selection managers to close this gap. Yann wrote:

Why not having this buffer in the X server directly?

Because it is not known ahead of time how big the selection data might be. Selections are often short text strings, but they can be anything including images, or arbitrarily large binary blob data. If the X server were to own this data the server would either have to start with a very large static data space - which still might occasionally be too small, or it'd have to dynamically allocate space as needed.

Far more importantly than this, though, is that X was designed as a network protocol. Under the original design the client and server were often not the same machine - and it still can be used this way even if many of us now have them both on the same physical machine. If the server were to store selection data, then anytime an image program copied image data to a selection, it'd have to send that data over the network. Then every time another client pasted the image, it would retrieve all that data over the network. Despite the sender and receiver being on the same physical machine, the data would have to be sent to another machine and back. This is very bad design. This is compounded by the fact that every time there was a copy operation, the data would have to be sent, whether or not any client would ever request it for pasting.

Contrast this with the design that was chosen: an image program copies large image data and all that is sent to the server is a very short message saying “I've copied something”. Then when another client program requests a paste, all the xserver sends back is the id of the selection owner. The pasting client contacts the copy client and the data is transferred directly from client to client on the same machine.

This was a very good design choice by the X11 developers.

https://github.com/zyedidia/micro/issues/538

Look at the program watch that can track changes in files, and run actions according to the user needs.

sudo touch /forcefsck; sudo reboot

Highlight a function name, then press Shift+k.

The manpager visualizer is by default less on my OS. But we can change it with export MANPAGER=nvim for example. However it does not display text and colors properly. We should find a nice solution.

Check if lsattr /the/file have a -i or -a. If yes : chattr -i /the/file or chattr -a /the/file. Then rm /the/file.

https://forums.docker.com/t/docker-daemon-error-invalid-character-x00-looking-for-beginning-of-value/18072/4

It appears this may be to invalid data written to /var/run/docker/libcontainerd/containerd/events.log when containerd shuts down improperly. We were able to move this file and restart docker, to recover our containers.

sudo dmesg -w

sudo memtester 8192 5

8192 = 8Go, 5 = 5 tests.

I zipped a folder under Windows to find it is broken when trying to unzip on Linux.

Doing this seemed to work :

zip -F the_broken.zip --out repaired.zip
#zip -FF the_broken.zip --out repaired.zip # only if the first line does not work
unzip repaired.zip

Inside /etc/pacman.conf :

# NOTE: You must run `pacman-key --init` before first using pacman; the local
# keyring can then be populated with the keys of all official Arch Linux
# packagers with `pacman-key --populate archlinux`.

about:processes

To enable dark theme in the last version (0.4) I had to execute this in a terminal :

gsettings set org.gnome.desktop.interface color-scheme 'prefer-dark'

Add this to .ssh/config:

Host 192.168.0.69
    Port 696969

It works! Else, we should change the remote in ALL repos… with:

git remote add origin ssh://git_user@git_server:git_port/PATH/TO/REPO

I discovered a tool rkhunter to track potential rootkits on a system.

rkhunter --update
rkhunter --propupd
rkhunter -c # to check

When I'm in a branch where I only want to merge a file from another branch but can't checkout until commiting change.

See : https://stackoverflow.com/a/18508527

git stash
git checkout <another-branch>
git stash apply
git add <one-file>
git commit 
git stash
git checkout <original-branch>
git stash apply

git stash pop

Will unstash the changes.

If you want to preserve the state of files (staged vs. working), use :

git stash apply --index

Example with ls

man -Tpdf ls > ls.pdf

I added the file 147-no_terminating_binding into /usr/share/X11/xorg.conf.d/ :

#/usr/share/X11/xorg.conf.d/147-no_terminating_binding
Section "ServerFlags"
        Option "DontZap"    "True"
EndSection

macOS runs by default an old version of bash (~3.2 apparently). The option -v has only been implemented in bash 4.2, so to convert new bash versions to old one for checking variables' existence :

-v     =    ! -z ${...+x}
! -v   =    -z ${...+x}

The idea is to copy the main branch, then remove it and rename. This will show only one commit in the new branch.

git checkout --orphan tmp-master # create a temporary branch
git add -A  # Add all files and commit them
git commit -m 'Add files'
git branch -D master # Deletes the master branch
git branch -m master # Rename the current branch to master
git push -f origin master # Force push master branch to Git server

https://unix.stackexchange.com/questions/28803/how-can-i-reduce-a-videos-size-with-ffmpeg

This answer was written in 2009. Since 2013 a video format much better than H.264 is widely available, namely H.265 (better in that it compresses more for the same quality, or gives higher quality for the same size). To use it, replace the libx264 codec with libx265, and push the compression lever further by increasing the CRF value — add, say, 4 or 6, since a reasonable range for H.265 may be 24 to 30. Note that lower CRF values correspond to higher bitrates, and hence produce higher quality videos.

ffmpeg -i input.mp4 -vcodec libx265 -crf 28 output.mp4

As of today, I don't manage to make it work using the GPU… So I use this command to convert from mkv (h264) to mp4 (h264_nvenc) :

ffmpeg -y -hwaccel cuda -hwaccel_output_format cuda -i input.mkv -c:v h264_nvenc output.mp4

⇒ 40% reduction (~11X speed) [ from 1300mo to 803 ]

I needed to try bash-3.2 that is the default shell on macOS. Thanks to Docker it's simple :

docker run -it bash:3.2

about:config ⇒ devp ⇒ layout.css.devPixelsPerPx ⇒ replace -1 with 1.5 (or less or more)

 wget --level=inf \
         --recursive \
         --page-requisites \
         --user-agent=Mozilla \
         --no-parent \
         --convert-links \
         --adjust-extension \
         --no-clobber \
         -e robots=off \
https://indulgent.website.url/

du -sh -- * | sort -h

mount /path/to/image.iso /mnt/iso -o loop

fc-list     # get list
fc-match sans # find matches with "sans"
fc-match monospace # find monospace fonts
fc-conflist

Config file: ~/.config/fontconfig/fonts.conf

Don't forget to change USER, SERVER_URL:

ssh USER@SERVER_URL "sudo -S dd if=/dev/mmcblk0 bs=64k status=progress | gzip -1 -" | dd of=rpiimg.gz bs=64k

ssh -C2qTnN -D 8080 USER@SERVER_URL

Then in the browser: create a SOCKS proxy with url 127.0.0.1:8080.

lpr file.txt -P PRINTER_NAME #tab to find printer