Rsync and encrypted SSH keys

Unfortunately rsync does not cooperate with $SSH_ASKPASS the way scp and ssh do.

Meanwhile, using scp is absolutely terrible on slow connections where files are being updated... To scp updates of a program to 4 servers (including one in China with the GFW) it was taking ~20 minutes. Running it with rsync where only deltas of the binary are sent, it happens in under 15 seconds.

So rsync is totally worth using.

Let's say that this is the rsync job that I want to run:

rsync --chown app:app -e 'ssh -p 2222' \
  --progress ./my-app \

All you need to do is add a -v to the part that calls ssh, and then grep for the Sending command line, like this:

rsync --chown app:app -e 'ssh -v -p 2222' \
  --progress /tmp/foo \
  app@MYSERVER:/home/app 2>&1 \
  | grep 'Sending command'

Enter passphrase for key '/home/me/.ssh/id': 
debug1: Sending command: rsync --server -oge.LsfxCIvu --log-format=X --usermap=\\*:app --groupmap=\\*:app . /home/app

The output of our rsync --server command is EXACTLY the only thing we'll allow in our ~/.ssh/authorized_keys on our server, and then we'll append:


So generate a new ssh key that we'll use just for rsyncing these files:

ssh-keygen -f rsync_key -t ed25519 -q -N ""

Then, on your server, edit the authorized keys file to add the content of But we'll prepend the command= parameters to restrict the key to this exact rsync command:

command="rsync --server -oge.LsfxCIvu  \
--log-format=X --usermap=\\*:app \
--groupmap=\\*:app . /home/app", \
no-pty,no-agent-forwarding,no-port-forwarding \
ssh-ed25519 AAAACFAKE_KEY_GENERATED_ONLY_FOR_TESTBOk7MpJi9jXfs+           
      ↳ 4lEOvpQFAKE_RSYNC \

Note that authorized keys lines must be on ONE LINE, so when you actually paste it in, it will look like:

command="rsync --server -oge.LsfxCIvu  --log-format=X --usermap=\\*:app --groupmap=\\*:app . /home/app", no-pty,no-agent-forwarding,no-port-forwarding ssh-ed25519 AAAACFAKE_KEY_GENERATED_ONLY_FOR_TESTBOk7MpJi9jXfs+           
      ↳ 4lEOvpQFAKE_RSYNC me@myhostname

How slow is Visual Studio Code?

Why does vscode feel so slow? Well, just look at the startup time:

❯ time code-oss && pkill -9 -f code-oss
Warning: 'app' is not in the list of known options, but still passed to Electron/Chromium.
'code-oss' time: 0.089s, cpu: 99%

❮ time nvim -c ":q"
'nvim -c ":q"' time: 0.025s, cpu: 31%

I used vscode for a while... Good programming language support is far more important than UI speed when using an editor.

VSCode did something amazing that had never been done before though: it developed an excellent standard for LSP, the Language Server Protocol, making it possible to connect these LSPs to other editors.

Thanks to vscode, it's now possible to get excellent programming language support into neovim!

Linux Windows

Sharing Windows OpenSSH keys for Linux Dual Boot

TL/DR: If you run into problems with opensshd permissions on windows, open a PowerShell Administrator prompt and run:

cd C:\ProgramData\ssh

takeown /R /F ssh_host*

icacls ssh_host* /T  /Q /C /RESET

icacls ssh_host* /grant SYSTEM:`(F`)

icacls ssh_host* /grant Administrators:`(F`)

icacls ssh_host* /inheritance:r

icacls ssh_host* /setowner system

Previously I wrote about Installing OpenSSH on Windows. For my workflow, I actually prefer to dual-boot Linux and Windows even though WSL2 has come a long way.

I use Barrier (open source successor to synergy) to share my mouse (well trackball) and keyboard across my workstation and laptop, regardless of whether Linux or Windows is running - I securely share the same underlying keys, and have the dhcp server assign a fixed IP to each MAC address.

It's actually quite tricky to get your OpenSSH keys from Linux's /etc/ssh/ssh_host_*key to C:\ProgramData\ssh\ssh_host_*key because of ACL details, even though I only edited the files with nvim - I thought that should preserve the icacls status, but it doesn't.

Windows iacls are a bit like selinux or AppArmor. Not a trivial subject, so be prepared if you're going to wade in.

iacls have inheritance, removed with /inheritance:r

For me, the most confusing thing about icacls is that if you break the permissions in certain ways (for example removing inheritance before you've granted some individual permissions to that file), you can no longer use icacls to fix them! You have to use takeown to re-assert ownership, and then you can start using icacls again.

PS C:\ProgramData\ssh> net stop sshd
The OpenSSH SSH Server service was stopped successfully.

PS C:\ProgramData\ssh> net start sshd
The OpenSSH SSH Server service is starting.
The OpenSSH SSH Server service could not be started.

A system error has occurred.
System error 1067 has occurred.
The process terminated unexpectedly.

Because I am not a master of icacls, I completely hosed my entire C:\ProgramData permissions while trying to fix ssh...

When trying to run sshd directly from the command line rather than via the windows service infrastructure, I actually got a bit more detail.

PS C:\WINDOWS\system32> sshd -dd
debug2: load_server_config: filename __PROGRAMDATA__\\ssh/sshd_config
debug2: load_server_config: done config len = 158
debug2: parse_server_config: config __PROGRAMDATA__\\ssh/sshd_config len 158
debug1: sshd version OpenSSH_for_Windows_8.1, LibreSSL 3.0.2
debug1: get_passwd: LookupAccountName() failed: 1332.
debug1: Unable to load host key: __PROGRAMDATA__\\ssh/ssh_host_rsa_key
debug1: Unable to load host key: __PROGRAMDATA__\\ssh/ssh_host_ecdsa_key
debug1: Unable to load host key: __PROGRAMDATA__\\ssh/ssh_host_ed25519_key
sshd: no hostkeys available -- exiting.

The PowerShell team provides a guide for exactly what ACL permissions are required to for your ssh_host_* files.

After several rounds of shooting myself in the foot with the not very memorable friendliness of icacls, I finally ran:

icacls "C:\ProgramData\ssh" /setowner system
icacls "C:\ProgramData\ssh" /q /c /t /reset
icacls "C:\ProgramData\ssh\ssh_host_*" /remove erwin

After that, I ran sshd -dd and finally was able to get OpenSSH to start up again on the command line without permissions errors, however running net start sshd still was failing to startup...

Turns out that running sshd -dd as just runs sshd in interactive mode under the currently logged on user (typically as an admin). To simulate the SYSTEM actually running sshd as a service you actually want to run:

psexec -s sshd.exe -ddd

(Note, psexec is part of sysinternals, probably easiest to install with shovel...)

PS C:\WINDOWS\system32> psexec -s sshd.exe -dd

PsExec v2.34 - Execute processes remotely
Copyright (C) 2001-2021 Mark Russinovich
Sysinternals -

debug2: load_server_config: filename PROGRAMDATA\ssh/sshd_config
debug2: load_server_config: done config len = 158
debug2: parse_server_config: config PROGRAMDATA\ssh/sshd_config len 158
debug1: sshd version OpenSSH_for_Windows_8.1, LibreSSL 3.0.2
debug1: get_passwd: LookupAccountName() failed: 1332.
Permissions for 'PROGRAMDATA\ssh/ssh_host_rsa_key' are too open.
It is required that your private key files are NOT accessible by others.
This private key will be ignored.
debug1: Unable to load host key "PROGRAMDATA\ssh/ssh_host_rsa_key": bad permissions
debug1: Unable to load host key: PROGRAMDATA\ssh/ssh_host_rsa_key
Permissions for 'PROGRAMDATA\ssh/ssh_host_ecdsa_key' are too open.
It is required that your private key files are NOT accessible by others.
This private key will be ignored.
debug1: Unable to load host key "PROGRAMDATA\ssh/ssh_host_ecdsa_key": bad permissions
debug1: Unable to load host key: PROGRAMDATA\ssh/ssh_host_ecdsa_key
Permissions for 'PROGRAMDATA\ssh/ssh_host_ed25519_key' are too open.

It is required that your private key files are NOT accessible by others.
This private key will be ignored.
debug1: Unable to load host key "PROGRAMDATA\ssh/ssh_host_ed25519_key": bad permissions
debug1: Unable to load host key: PROGRAMDATA\ssh/ssh_host_ed25519_key
sshd: no hostkeys available -- exiting.
sshd.exe exited on XPS with error code 1.

So even though I had fixed the permissions enough for my user to run sshd it was not enough for the system to run sshd, the way that net start sshd works.

I wasn't even able to cd C:\ProgramData\ssh so I started with:

get-acl C: | set-acl C:\ProgramData

Then when I cd into C:\ProgramData\ssh turns out that the permissions are in fact way more open than what Window's SSHD (or Linux sshd for that matter) permit.

PS C:\ProgramData\ssh> icacls ssh_host_*key
ssh_host_dsa_key NT AUTHORITY\Authenticated Users:(I)(M)
                 NT AUTHORITY\SYSTEM:(I)(F)

ssh_host_ecdsa_key NT AUTHORITY\Authenticated Users:(I)(M)
                   NT AUTHORITY\SYSTEM:(I)(F)

ssh_host_ed25519_key NT AUTHORITY\Authenticated Users:(I)(M)
                     NT AUTHORITY\SYSTEM:(I)(F)

ssh_host_rsa_key NT AUTHORITY\Authenticated Users:(I)(M)
                 NT AUTHORITY\SYSTEM:(I)(F)

Successfully processed 4 files; Failed processing 0 files

So the easy way to do this on Windows is just to focus on one file at a time... We know that sshd complained about ssh_host_rsa_key first, so we'll start there.

icacls .\ssh_host_rsa_key /inheritance:r

So this removed the inheritance ACL from that file, and it's basically impossible to re-add...

Windows improbable answer for removing inheritance from a single file is to use takeown

takeown /R /F C:\ProgramData\ssh

# Then reset the ACLs to their default values

icacls C:\ProgramData\ssh /T /Q /C /RESET

After takeown runs, you'll be able to fix all the permissions again, but all the permissions will be messed up requiring them to be fixed 😉

Now we'll try again:

icacls.exe .\ssh_host_rsa_key

.\ssh_host_rsa_key NT AUTHORITY\Authenticated Users:(I)(M)
                   NT AUTHORITY\SYSTEM:(I)(F)

Now I'm first going to explicitly grant Full Control to System.

icacls.exe .\ssh_host_rsa_key /grant SYSTEM:`(F`)
icacls.exe .\ssh_host_rsa_key /grant Administrators:`(F`)

Then I'm going to remove Inheritance

icacls.exe .\ssh_host_rsa_key /inheritance:r

However, unfortunately this still doesn't work... Even though this gives us the exact values that the PowerShell team documents as required:

ssh_host_rsa_key BUILTIN\Administrators:(F)

When starting up sshd with psexec we'll still get the error:

Permissions for '__PROGRAMDATA__\\ssh/ssh_host_rsa_key' are too open.
It is required that your private key files are NOT accessible by others.
This private key will be ignored.
debug1: Unable to load host key "__PROGRAMDATA__\\ssh/ssh_host_rsa_key": bad permissions
debug1: Unable to load host key: __PROGRAMDATA__\\ssh/ssh_host_rsa_key

After far too long of a detour today, I finally solved it with:

cd C:\ProgramData\ssh

takeown /R /F ssh_host*

icacls ssh_host* /T  /Q /C /RESET

icacls ssh_host* /grant SYSTEM:`(F`)

icacls ssh_host* /grant Administrators:`(F`)

icacls ssh_host* /inheritance:r

icacls ssh_host* /setowner system

Unlike Linux, the parent directory permissions don't seem to matter.

Now, net start sshd works perfectly 🙂


How to completely remove Snaps? Install voidlinux

Ubuntu Linux is on a dark path these days. I used Debian on a lot of servers in the early 2000s, and like RedHat Enterprise, the Debian Stable stuff is just too old... So having a team "stabilize" Debian's Unstable distribution every six months was a fantastic idea and it worked well for a long time.

Unfortunately there is a big disconnect between who the Ubuntu team wants to serve and it's regular users. Sadly, Wayland seems to have this same disconnect, but that's another rant.

Snaps are just so awful. They break your themes, break your plugins and thus your workflow, block your access to files, clutter up your process list and mount list. Containerization is a great idea AS LONG AS you don't reduce function... So far, containerization on the desktop is not working. I would much rather use a heaver QubesOS like model rather than kind of working but breaking tons of advanced use cases.

voidlinux is an awesome alternative to Ubuntu:

  • Gets rid of Systemd, using very simple runit scripts for services
  • Void is not a FORK of anything. It's a new Linux developed from scratch.
  • It's a "Stable rolling release" distro, while Arch just focuses on Bleeding Edge rolling releases (that occasionally blow up in my face)
  • Easily install any Debian package with xdeb
    • I used it for Zoom, Opera and Jumpapp
  • Enjoy the super simple xbps package system
    • sudo xbps-install -Svu to update the cache the first time
    • sudo xbps-install etckeeper to install etckeeper
    • sudo xbps-install xtools to install xlocate
  • xlocate is a very fast replacement for apt-file search, just type xlocate '/xlocate$' it accepts any regex and will help you find any file inside of any xbps package, even the ones you haven't installed.
  • void is a minimal distro without any bloat

To get the install started, you just download the voidlinux ISO, which will startup a voidlinux live image. Open the terminal and run sudo void-installer - the password is "voidlinux".

That's basically it. Once you're booted into your own voidlinux install, install xtools and run xlocate to find whatever you're missing. Here's my list of daily use tools if you need something to help get started.

# Refresh the XBPS-Cache
sudo xbps-install -Svu

# Always start with etckeeper
sudo xbps-install etckeeper

# Install generic utilities
sudo xbps-install htop neofetch python3 sqlite kitty kitty-terminfo 
sudo xbps-install ddcutil mlocate ntp
sudo xbps-install tree ranger tmux screen direnv mosh fzf fzy
sudo xbps-install bind-utils git curl wget whois netcat socat
sudo xbps-install flameshot zsh xtools syncthing cheat sxhkd
sudo xbps-install rofi aws-cli zathura tdrop mpg123 mp3info 
sudo xbps-install xsel xmodmap xev xrandr wmctrl jump
sudo xbps-install binutils tar curl xbps xz bash-completion
sudo xbps-install base-devel

# KDE Plasma Desktop
sudo xbps-install plasma-desktop
sudo xbps-install kde5-baseapps kde-cli-tools qt5-tools
sudo xbps-install kdegraphics-thumbnailers ffmpegthumbs kwallet-pam
sudo xbps-install ksshaskpass kwalletmanager 
sudo xbps-install pavucontrol-qt pavucontrol

Linux X11

Adding a `hyper` modifier key to your keyboard, because shift/ctrl/alt/meta/super aren’t enough

TL/DR: Try this in your ~/.Xmodmap log out of X11 and log back in. If you moved to Wayland, hopefully one day we'll get stable APIs for 3rd party window utilities. Until that happens, I recommend switching back to X11 if you can.

clear mod1
clear mod2
clear mod3
clear mod4
clear mod5

keycode 248 = Hyper_L

add Mod1 = Alt_L Alt_R
add Mod2 = Num_Lock
add Mod3 = Hyper_L
add mod4 = Super_L Super_R
add mod5 = ISO_Level3_Shift Mode_switch

Note that you can't blindly copy code 248... You need to find a key that will work for your keyboard.

Soft fn key on keyboard

Use xev -event keyboard to find the keycode for the key that you want to use. While xev is running, focus on the xev window and press the key you want to use. You'll see the KeyRelease event show on your console. Best to use the xev -event keyboard filter so that all mouse events can be ignored.

In my example, this is AFTER I've use Xmodmap to map this key to Hyper_L... You won't see this until your remapping is successful.

You need to make sure Xmodmap will actually start with your widow manager, and for me, it works best with a delay. So I have a file ~/.config/autostart/xmodmap.desktop

[Desktop Entry]
Comment=Xmodmap Keyboard Modification

And then delay_xmodmap is a trivial script that just sleeps for a bit before launching.


sleep 10;
xmodmap ~/.Xmodmap

Using a hot key manager like sxhkd you should be able to map shortcuts to the Hyper key now. Put something like this in your ~/.config/sxhkd/sxhkdrc to test that Hyper is working.

hyper + {a-z}
        notify-send "hyper {a-z}"

Every keyboard I've ever seen has a SHIFT key and CONTROL keys. Historical Unix vs Mac vs Windows are a bit different for the other keys:
Alt == Meta (Option in the Mac World)
Windows Key == Super Key == Command Key

So you basically get 3 modifiers on each key, but SHIFT is so common for text it's dangerous for most of us to configure that.

If you're on Linux, and you have a keyboard that generates an actual "keycode" (some fn labeled keys actually run software in the keyboard to send a different keycode to the OS, without the OS knowing that a modifier was involved).

mod1AltOptionAlt_L Alt_R / Meta
mod2Num LockNum Lock
mod4Windows KeyCommand KeySuper_L Super_R
X11 mod1 keys


Git Self Hosted: GitLab vs Gitea vs Gogs in 2022

Github is a double edged sword. Use it when you must, but you should self-host whenever possible.

I started with gitlab as it was the most popular github alternative at the time, and I think for medium-large businesses hosting their own local, dedicated git instance, gitlab is probably a good choice.

For me though, gitlab wad definitely overkill. The upgrades get very complicated, and I'm ultimately just not a big Ruby on Rails fan... I just wanted locally hosted git that I could also browse easily via http, especially to graphically view older commits and branches, because every once in a while I do leave the terminal...

After gitlab my next solution was Gitea, as I read it was a community fork of gogs so I assumed it must be better... In hindsight, I would say Gitea is working hard to compete with GitLab. That's fine, but `GitLab` is actually what I wanted to get away from.

You can find a Gitea feature list here:

Here's an old sysadmin rule of thumb: If it involves LDAP, run the other way!

Me, screaming... every time I do something with LDAP...

In all fairness, LDAP is necessary for directory based authorization and authentication. With lots of users, you're kind of stuck with it and so you better know how it works. Everything connected to X.509 is a whole different mindset than the rest of the Linux and Open Source movement. If software had an attitude, ldap software and the zshell would be at opposite edges of the attitude spectrum.

I spent a long time actually getting gitea to work in my particular configuration, and just ended up dedicating way more time to it than I really wanted to. For me, git isn't very much different from samba... I just want it to work so I can focus on problems that I'm more interested in.

I would recommend gitea to anyone looking at gitlab and vice versa. These two are very similar. I think that blindly running either of them in a container just because you can is asking for trouble though. Go through the manual instillation and know how to set things up from scratch. If you can't do that, you shouldn't run it, because you won't be able to fix it when things go wrong. You want a sysadmin that knows how to set these up and how to manage them, back them up, and fix problems along the way.

So I spent far more hours than I'm proud to admit trying to get gitea working in my particular environment, but when I decided to switch over to gogs, I had it all working and my old repos migrated over within 30 minutes.

If you're just looking for a simple, self hosted git server via `https` in 2022, I think that gogs is what you are looking for.

Another nice "self hosted" option is cgit but the default interface is very dated, and it's even more minimal than my ideal feature set.

If neither gitlab, gitea, gogs or cgit are what you're looking for, maybe check out the awesome self hosted list, but a lot of these are very niche and not well supported.


kitty and Nerd Fonts… bringing Egyptian Hieroglyphs to your terminal…

TL/DR: Just download Nerd Symbols Only Font and Config Kitty

If you haven't tried the kitty terminal emulator, it's an awesome one. Even if I was still tortured to use Mac OS X, I think I would use Kitty of iTerm. You'll find that it's noticeably faster than gnome-terminal or vscode's built in terminal. (konsole is pretty snappy though)

Kitty has tons of features, however the process to discover and configure those features is a bit more complicated than other terminals that I've used.

For me, the most confusing aspect of kitty was understanding how to setup fonts. I never wanted to learn how kitty deals with fonts. I just wanted a kitty font config that works. If that's all you want, skip to kitty config.

First, Kitty requires that fonts have a fixed spacing of 100 defined. To check if a font can be used by Kitty, and the exact name you should enter to configure it, you'll want to use:

kitty list-fonts

If you're looking for a good font to use, I suggest you take a look at the open source Fira Code font, since it has nice ligatures for programming...

Fira Code example "ligatures" o visually simplify display of == --> ++ and :=

If you you have a budget to invest in a programming font, I think the MonoLisa font is even more attractive. To get a monospaced font to show up in Kitty if it's not showing up on it's own, you'll need to edit ~/.config/fontconfig/fonts.conf and add a section to "scan" for the font "family" and set "spacing" for that family to 100.

<?xml version='1.0'?>
<!DOCTYPE fontconfig SYSTEM 'fonts.dtd'>

<match target="scan">
    <test name="family">
    <edit name="spacing">


Once you edit fontconfig/fonts.conf you'll need to rebuild the font cache and restart kitty to see the new font.

sudo fc-cache -fr

If kitty list-fonts still can't find the font that you're looking for, several steps possible steps to resolve are discussed on the kitty issue tracker.

Most terminals need you to setup "patched fonts" if you want to use "powerline" symbols... If you're looking to rice your command line, powerlevel10k, vim-devicons, and vim-airline can will use many of these "patched font" symbols.

vim-airline using patched fonts...
Powerlevel10k Extravagant Style
powerlevel10k zsh theme using patched fonts
vim-devicons overall screenshot
vim-devicons with patched fonts

If you want to use the "fontconfig" method for your whole configuration, you might be able to add something like the following ~/.config/fontconfig/fonts.conf so that symbols from the Symbols Nerd Font will be preferred, but I was not able to get this method to work, so I went with the kitty symbol_map method.

   <family>Symbols Nerd Font</family>
   <family>Noto Color Emoji</family>
   <family>Noto Sans Symbols2</family>

Yes, you list all of them them [symbol_map] and yes you can have multiple ones. However, I recommend against doing that, there is rarely a need for it, since nerd fonts will automatically be used if found in most cases.

kovidgoyal, author of kitty

For me though, "auto detection" of nerd fonts never worked. The most evident example was each time I would run:

vim ~/.vimrc

The stylized "V" from vim-devicons would show a very rarely used Chinese character...

Editing .vimrc, the stylized V is shown for me as: 金 + 弦, which doesn't exist in most chinese dictionaries...
Editing .vimrc, the correct stylized "V" icon from vim-devicons

The "quick and dirty" way to get every symbol you want to load is the use the patched Nerd fonts, for example the patched Nerd Fonts.

However, with Kitty, the preferred way to get each font to show up is actually NOT to use patched fonts, instead to use kitty's "symbol_map" function.

Map the specified unicode codepoints to a particular font. Useful if you need special rendering for some symbols, such as for Powerline. Avoids the need for patched fonts. Each unicode code point is specified in the form U+<code point in hexadecimal>. You can specify multiple code points, separated by commas and ranges separated by hyphens. symbol_map itself can be specified multiple times. Syntax is:

symbol_map U+E0A0-U+E0A3,U+E0C0-U+E0C7 PowerlineSymbols
symbol_map codepoints Font Family Name

If you run kitty with, you'll get a list of each character code that's not found on the terminal you launched kitty from.

kitty --debug-font-fallback

For example, on the Right I'm using xterm to edit ~/.config/kitty/kitty.conf and on the left is my kitty instance... Here you can see pretty clearly the "DejaVu Sans Mono" font is providing the U+2605 (★) and U+21b6 (↶)

I would prefer to get all of the Nerd Fonts from the Nerd Fonts package.

First, make sure you download the "Symbols Nerd Font" (symbols only) font package, you can find it at:

src/glyphs/Symbols-2048-em Nerd Font Complete.ttf

Once you have the Symbols-2048-em Nerd Font Complete.ttf, copy it to ~/.local/share/fonts and then run sudo fc-cache -fr to reload the font cache.

Next, you'll want to add something like this to your kitty config:

font_family FiraCode
# or #
font_family MonoLisa

# Symbols Nerd Font complete symbol_map
# easily troubleshoot missing/incorrect characters with:
#   kitty --debug-font-fallback

# "Nerd Fonts - Pomicons"
symbol_map  U+E000-U+E00D Symbols Nerd Font

# "Nerd Fonts - Powerline"
symbol_map U+e0a0-U+e0a2,U+e0b0-U+e0b3 Symbols Nerd Font

# "Nerd Fonts - Powerline Extra"
symbol_map U+e0a3-U+e0a3,U+e0b4-U+e0c8,U+e0cc-U+e0d2,U+e0d4-U+e0d4 Symbols Nerd Font

# "Nerd Fonts - Symbols original"
symbol_map U+e5fa-U+e62b Symbols Nerd Font

# "Nerd Fonts - Devicons"
symbol_map U+e700-U+e7c5 Symbols Nerd Font

# "Nerd Fonts - Font awesome"
symbol_map U+f000-U+f2e0 Symbols Nerd Font

# "Nerd Fonts - Font awesome extension"
symbol_map U+e200-U+e2a9 Symbols Nerd Font

# "Nerd Fonts - Octicons"
symbol_map U+f400-U+f4a8,U+2665-U+2665,U+26A1-U+26A1,U+f27c-U+f27c Symbols Nerd Font

# "Nerd Fonts - Font Linux"
symbol_map U+F300-U+F313 Symbols Nerd Font

#  Nerd Fonts - Font Power Symbols"
symbol_map U+23fb-U+23fe,U+2b58-U+2b58 Symbols Nerd Font

#  "Nerd Fonts - Material Design Icons"
symbol_map U+f500-U+fd46 Symbols Nerd Font

# "Nerd Fonts - Weather Icons"
symbol_map U+e300-U+e3eb Symbols Nerd Font

# Misc Code Point Fixes
symbol_map U+21B5,U+25B8,U+2605,U+2630,U+2632,U+2714,U+E0A3,U+E615,U+E62B Symbols Nerd Font

The list of code point ranges above comes directly from the nerd-fonts test script, with the exception of the "Misc Code Point Fixes" that I added myself through trial and error with --debug-font-fallback.

You can test that you've got all fonts configured properly by running the "" script provided by nerd-fonts.

Be sure to run kitty with kitty --debug-font-fallback then run this script in your kitty window, and if you've set everything up correctly, you should not see any of the symbol missing or symbol fallback notices that were shown in the screenshot above. Once properly configured you'll get:

Note that there may still be a few messages, but generally all characters will now load correctly.
Small excerpt of the "" script once kitty is properly configured

If you're experimenting with MonoLisa, here are some settings that I found looked great.

font_family      MonoLisa
font_bold        MonoLisa-Bold
italic_font      MonoLisa-Light
adjust_baseline -4
font_size        14

Without the "adjust_baseline -4", the MonoLisa font was not aligned when I tried.

One last nugget of wisdom here, the "" site is the best unicode character lookup site I've ever seen. You can type the codepoint directly in the URL, and the layout is clean and free of ads.

Did you know there are Unicode codepoints for Egyptian Hieroglyphs?
Now you can put heiroglyphs in your terminal!"

Egyptian Hieroglyphs in markdown document displayed via kitty terminal
# Egyptian Hieroglyphs are Back!

  𓀀 𓀁 𓀂 𓀃 𓀄 𓀅 𓀆 𓀇 𓀈	𓀉 𓀊 𓀋 𓀌 𓀍 𓀎 𓀏 𓀐 𓀑
    𓀒 𓀓 𓀔 𓀕 𓀖 𓀗 𓀘 𓀙 𓀚	𓀛 𓀜 𓀝 𓀞 𓀟 𓀠 𓀡 𓀢 𓀣
    𓀤 𓀥 𓀦 𓀧 𓀨 𓀩 𓀪 𓀫 𓀬	𓀭 𓀮 𓀯 𓀰 𓀱 𓀲 𓀳 𓀴 𓀵
    𓀶 𓀷 𓀸 𓀹 𓀺 𓀻 𓀼 𓀽 𓀾	𓀿 𓁀 𓁁 𓁂 𓁃 𓁄 𓁅 𓁆 𓁇
    𓁈 𓁉 𓁊 𓁋 𓁌 𓁍 𓁎 𓁏 𓁐	𓁑 𓁒 𓁓 𓁔 𓁕 𓁖 𓁗 𓁘 𓁙
    𓁚 𓁛 𓁜 𓁝 𓁞 𓁟 𓁠 𓁡 𓁢	𓁣 𓁤 𓁥 𓁦 𓁧 𓁨 𓁩 𓁪 𓁫
    𓁬 𓁭 𓁮 𓁯 𓁰 𓁱 𓁲 𓁳 𓁴	𓁵 𓁶 𓁷 𓁸 𓁹 𓁺 𓁻 𓁼 𓁽
    𓁾 𓁿 𓂀 𓂁 𓂂 𓂃 𓂄 𓂅 𓂆	𓂇 𓂈 𓂉 𓂊 𓂋 𓂌 𓂍 𓂎 𓂏
    𓂐 𓂑 𓂒 𓂓 𓂔 𓂕 𓂖 𓂗 𓂘	𓂙 𓂚 𓂛 𓂜 𓂝 𓂞 𓂟 𓂠 𓂡
    𓂢 𓂣 𓂤 𓂥 𓂦 𓂧 𓂨 𓂩 𓂪	𓂫 𓂬 𓂭 𓂮 𓂯 𓂰 𓂱 𓂲 𓂳
    𓂴 𓂵 𓂶 𓂷 𓂸 𓂹 𓂺 𓂻 𓂼	𓂽 𓂾 𓂿 𓃀 𓃁 𓃂 𓃃 𓃄 𓃅
    𓃆 𓃇 𓃈 𓃉 𓃊 𓃋 𓃌 𓃍 𓃎	𓃏 𓃐 𓃑 𓃒 𓃓 𓃔 𓃕 𓃖 𓃗
    𓃘 𓃙 𓃚 𓃛 𓃜 𓃝 𓃞 𓃟 𓃠	𓃡 𓃢 𓃣 𓃤 𓃥 𓃦 𓃧 𓃨 𓃩
    𓃪 𓃫 𓃬 𓃭 𓃮 𓃯 𓃰 𓃱 𓃲	𓃳 𓃴 𓃵 𓃶 𓃷 𓃸 𓃹 𓃺 𓃻
    𓃼 𓃽 𓃾 𓃿

Customizing Firefox shortcut keys… Get rid of CTRL + SHIFT + C devtools inspector shortcut

Firefox used to have some plugins that you could use to customize your command keys pretty easily, but as Firefox has evolved recently, many of the XUL generation of shortcut keys has been disabled.

It's very easy to rebuild Firefox to support the keys that you want though. First, you'll want to run once to build the initial "" file.

On Linux, the relative file path is:


Open, you'll find it in:
Then you'll want to change the default C shortcut key that conflicts with my terminal muscle memory of CTRL+C to interrupt, and Shift + Ctrl + C as Copy...

# LOCALIZATION NOTE (inspector.commandkey):
# Key pressed to open a toolbox with the inspector panel selected

The full workflow is...

# Grab the friendly mozilla build bootstrapper
cd ~/SourceInstall

curl -O

# Run the bootstrap script

# Do a test build to make sure it works
./mach build

./mach run

# edit file you want to change

./mach build

./mach run

# test your shortcut now works...

cd ~/SourceInstall/mozilla-unified/obj-x86_64-pc-linux-gnu/dist/bin

# or link to anywhere else in your path that
# will have higher precedence than /usr/bin

sudo ln -s $HOME/SourceInstall/mozilla-unified/obj-x86_64-pc-linux-gnu/dist/bin/firefox /usr/local/bin/firefox

Enjoy beginning to really customize Firefox... Hopefully for you this is the beginning of more meaningful Firefox hacking than just editing a config file.