Rudimentary time or counter based mechanisms are okay, until a more reliable or per-URL conditional can be thought of.

If you add 'update hostname if you're planning to' to the preparation section, that'd be enough.

The README is practically blank!

Bruh... what have I become! 🤦

Especially when updating/reinstalling.

Easy Mode:

1. set pi to book from USB
2. hold shift to start the network installer
3. When booted, plug in USB stick
4. choose 32bit
5. modify the settings before apply
6. success

Seems like this would be a good option here:

https://raspberrypi.stackexchange.com/questions/145382/remove-hide-mouse-cursor-when-idle-on-rasbperry-pi-os-bookworm/145390#145390

2

I've had success with installing the hideaway plugin for Interception Tools using the script below.

Adapted from https://forums.raspberrypi.com/viewtopic.php?t=358285#p2176499

#!/bin/bash
# hide mouse in wayland raspbian

sudo apt install -y interception-tools interception-tools-compat
sudo apt install -y cmake
cd ~
git clone https://gitlab.com/interception/linux/plugins/hideaway.git
cd hideaway
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build
sudo cp /home/$USER/hideaway/build/hideaway /usr/bin
sudo chmod +x /usr/bin/hideaway

cd ~
wget https://raw.githubusercontent.com/ugotapi/wayland-pagepi/main/config.yaml
sudo cp /home/$USER/config.yaml /etc/interception/udevmon.d/config.yaml
sudo systemctl restart udevmon

Make sure to back up the current config before cleaning up.
Update setup script to use the backed up config if available.

Blocks #9

The GUI needs to:

• Not be a total disastrous misnomer of aesthetic
• Display list of URLs (used as PiOSK screens)
• Have UX for adding & removing URLs
• Have UX for applying the changes
• Show footer with metadata

Send process to background.

This would be super handy! BTW, I love this.

Currently we're using nodejs to run the backend service for the web GUI. There are a couple of inconveniences to this:

• Lack of first party debian aptitude package
• Best effort (but still clumsy) third party installation script
• Necessitates dependency installation during setup process
• Which necessitates dependency version locking management
• Because entrypoint script takes argument (node index), it needs to be wrapped in a script to be used as ExecStart command

I think this can be much more effective, efficient and elegant by using Node.js SEA. We can use GitHub actions to generate the package, and in future, instead of cloning the entire repo, hopefully we'll just need to download the release binary.

Add update and uninstall docs.

Update major version due to breaking changes between the new setup and cleanup script having breaking changes compared to previous approach.

Thankfully we can still to do that before too many users start using it. Let's make sure people start with v2 - so that at least there's a cleanup script.

Releasing v1 without cleanup or update script was plain stupid, with severe lack of foresight.

Just thought I'd highlight that you can run local content.

Don't wait for the rc script (next reboot) to start dashboard server.

Allow the user to update config BEFORE the reboot.

This utility is meant to be "set it and forget it", so no one is forced to update if they're satisfied with how it's working for them. It shouldn't even have to have automated updates, notifying annoying them to update.

But at least for the ones who want to stay at the bleeding edge & update, it shouldn't be more than just one or two scripts to run.

The frontend needs to

• Be able to fetch & parse list of URLs
• Generate elements for URL list from HTML templates
• Serialize the updated list of URLs and POST request to update config

Fake API or assume defaults until backend is ready.

The backend needs to:

• Passthrough serve static assets
• Read config file, serialize data & send to frontend
• Accept new data, update config, reboot system to apply it

Simplify starting up dashboard server by using /etc/rc.local

The setup script needs to:

• install dependencies: wtype, git, nodejs,
• clone this repo in ~/ directory
• setup kiosk mode autostart
• setup web service daemon

Basically if the user performs the following steps, the rest should be handled by PiOSK after that:

1. flash OS image, boot up the SBC and ssh into it
2. update software packages
3. enable autologin
4. run this setup script

• Abstract the wayfire autostart execution script (to avoid touching it more than once).
• Read the config file for list of URLs
• Open up chrome with the list of URLs
• Use wtype scripts to swtich browser tabs

If an backup config exists (from prior uninstallation or otherwise), use that instead of sample config).

• Allows using port 80
• Allows rebooting system
• Avoids hard dependency on CLI
• Remains hands-free even on reset