Eser DENIZ / @SRWieZ

Linux suspend on a T2 MacBook Pro: what actually worked

2026-07-15T00:00:00+00:00

I recently installed CachyOS on my 2020 Intel MacBook Pro. Almost everything worked surprisingly well. Then I closed the lid.

The Mac went to sleep, the screen turned black, and it never came back. The Touch Bar sometimes lit up, but the display, keyboard, trackpad and network stayed dead. The only option was holding the power button and forcing a reboot.

This MacBook is a MacBookPro16,2 with Apple's T2 chip. Linux support is much better than it used to be, but suspend is still complicated.

What I tried
Trying t2bce
Hibernate did not save me
What actually works
The conclusion

What I tried

I tested both sleep modes available on the machine:

deep
s2idle

Both failed.

I then tried the older T2Linux workaround that unloads apple_bce before sleep and reloads it after resume. It did not help.

My CachyOS kernel already included the newer apple_bce no-state suspend implementation, so I removed the old workaround and tested the native path with pm_async=off. This was promising because other MacBookPro16,2 users reported successful resume with it.

Mine still froze.

Trying t2bce

The most interesting modern option is t2bce, a new replacement for the old apple_bce driver.

I built version 0.06 directly against the CachyOS kernel. Because CachyOS uses a Clang/LTO kernel, I had to build it with:

Copied!

make LLVM=1

I installed its four modules and created a one-time GRUB entry that disabled apple_bce. This gave me a safe fallback if the experimental boot failed.

The result was impressive at first. CachyOS booted normally and t2bce restored everything:

Internal keyboard and trackpad
Touch Bar
Audio
Wi-Fi

Unfortunately, real deep suspend still failed to resume and required another forced reboot.

Hibernate did not save me

My next idea was hibernation.

I created a dedicated 20 GB Btrfs swapfile, configured the correct resume offset, and verified that systemd could detect it. The Mac went dark, but did not wake normally. After a forced reboot, Linux reported that no valid hibernation image could be restored.

I also tried systemd's test_resume mode. That froze too.

At that point, suspend and hibernate had both exhausted enough of my patience.

What actually works

I stopped trying to make the Mac sleep.

My final lid configuration is simple:

Closing the lid locks KDE
The display turns off after five seconds
Automatic suspend is disabled
Opening the lid immediately shows the locked session

I also added a systemd-logind fallback:

Copied!

[Login]
HandleLidSwitch=lock
HandleLidSwitchExternalPower=lock
HandleLidSwitchDocked=ignore

Plasma's PowerDevil normally handles the lid event, but this ensures that logind never falls back to suspend when the desktop session is unavailable.

It is not real sleep, so it still uses battery. But it is predictable, instant and does not corrupt my session with forced shutdowns.

The conclusion

Linux on a T2 MacBook is now good enough to use every day, but sleep remains very dependent on the exact model, firmware and kernel.

t2bce is promising and I will keep watching it. For now, the reliable solution on my MacBookPro16,2 is not another suspend patch. It is accepting that the machine should stay awake when the lid is closed.

Not perfect, but it works every time.

Is Starlink Worth It? My Switch to Better Alternatives

2026-05-25T00:00:00+00:00

I had Starlink on the van for a while. It was great. I just cancelled it and went back to a SIM card in a 5G router. Here is why.

Life before Starlink
Why I tried Starlink
The problems piled up
My savior: Forfait Free Max — 20€/month
Bonus trick: Devolo Magic 2

Life before Starlink

Before Starlink, I had a 4G router. I travelled with it all the way to Turkey, Norway, Portugal, and every country in between. Full coverage almost everywhere.

It only ever failed me in rare spots: on top of a mountain right under the antenna, or stuck between two valleys with nothing in line of sight.

So why did I switch in the first place? Data.

I had to be really thoughtful about my usage. I disabled every auto-update I could find, checked my consumption every single day, and ended up going to cafés and laundromats whenever I needed to push a real update to my devices.

That is the part that wore me down.

Why I tried Starlink

I met a few people on the road who had it and it was flawlessly good. Fast, stable, no data limit. Exactly what I wanted: stop thinking about megabytes.

And honestly, it delivered.

The good:

300 Mbps down, 30 Mbps up — way more than I need.
Works everywhere in Europe I have been so far.
I can game, download large files, and livestream without worrying about it.

So what is the problem?

The problems piled up

The longer I used it, the more the downsides started to weigh on me.

I can't park under a tree anymore. The antenna needs a clear view of the sky, which is very annoying in the summer.
One more thing to orient. I already think about sun direction and how level the car is parked. Now I also have to think about where the antenna points.
Energy hungry. It pulls a continuous 40W. My old 5G router pulled 5W. On a van battery setup, that is a huge difference. I wanted to automate a shutdown at night but that's when my iphone syncronize everything.
95€/month, and going up again. That is the second price increase. At some point you have to ask yourself what you are actually paying for.

My savior: Forfait Free Max — 20€/month

A French mobile plan with unlimited data and proper EU roaming, dropped into a 5G router. That is it.

Back to ~5W of consumption
Works under trees
75€ less every month

And honestly it's as fast as before and even more on the uplaod side.

Bonus trick: Devolo Magic 2

A little extra setup I am pretty proud of.

I added a powerline adapter (CPL) inside the van, wired between my exterior shore power inlet and my Victron inverter. The second adapter lives in a small bag I keep in a drawer.

When I park at a friend's place and plug the van into their outdoor socket, I hand them the second adapter and ask them to plug it next to their router. That is it. The van is now on their home network through the power cable.

No coverage to worry about, no data usage, no Wi-Fi password to type on every device. And it works at full LAN speed.

I made 4 plugins for NativePHP

2026-04-16T00:00:00+00:00

I've been involved in one way or another with the NativePHP team since the rise of the desktop version. I've followed their journey from the backstage and helped them when I could. I'm extremely proud of what they have accomplished until now.

When Shane and Simon introduced me to the plugin system, I knew I wanted to explore its full potential.

Let me tell you, there are no limits to what you can achieve.

I created four basic plugins to meet my needs and decided to publish them.

But I assure you, you can run any native code you want with NativePHP; it's fast and reliable.

The plugins
Why paid?
What's next

The plugins

nativephp-mobile-screen (free)

A small plugin to keep the screen awake and control brightness from your NativePHP mobile app. Useful for anything that stays on display: dashboards, readers, kiosk-style apps.

Grab it on GitHub.

nativephp-mobile-contacts (paid)

Read and write device contacts straight from PHP. Handy when your app needs to pick a recipient, sync with the address book, or add a new contact after an in-app action.

Available on the NativePHP marketplace.

nativephp-mobile-calendar (paid)

Create, update and read calendar events from your NativePHP app. Great for booking flows, reminders, or any feature that needs to land on the user's calendar without leaving your app.

Available on the NativePHP marketplace.

nativephp-mobile-screenshots (paid)

Capture the current screen programmatically — for in-app bug reports, share sheets, or "save this view" features.

Available on the NativePHP marketplace.

Why paid?

The three mobile plugins above ship with native Swift and Kotlin code I had to test and maintain on both platforms. Selling them on the marketplace is what makes it sustainable to keep them up to date as iOS and Android evolve.

nativephp-mobile-screen stays free and open source — it's a good starting point if you want to see how a NativePHP mobile plugin is structured.

What's next

I have a few more ideas, but I want to see if those will sell first.

Maintaining code is a full-time job, and I want to be compensated for it.

If those three are successful enough, I will likely continue creating new plugins.

At the very least, I will develop one more that I need for an app.

Devkeepr: your trusted companion for your dev environment

2026-04-14T00:00:00+00:00

Count your node_modules folders. Now multiply by 400MB. Yeah, that's the problem Willem and I have been quietly fixing for the past few months.

What's Devkeepr?
What it does
Why we built it
Try it

What's Devkeepr?

Devkeepr is a desktop app that takes care of your dev machine while you're not looking. Think CleanMyMac, but built by people who actually know what vendor, node_modules, .venv, target and bin/Debug are doing to your SSD.

It detects 50+ stacks (PHP, Node, Python, Ruby, Go, Rust, .NET…) and gives you a single command center for all your projects.

What it does

Smart hibernation — Idle projects get their dependencies, caches and build artifacts wiped automatically. They wake up the moment you cd into them or open them in your IDE. A 2.4 GB project becomes 3.8 MB on disk until you actually need it again.
Project command center — Run npm scripts, composer scripts, Makefile targets or GitHub workflows from one place, across all your projects.
Global updater — Keeps your global packages up to date on a schedule.
Disk usage dashboard — See exactly where your disk space is going across every project.
Git auto-sync — Fetches and prunes remote branches so your repos never go stale.
Workflow & deployment monitors (Pro) — Watch CI/CD on GitHub, GitLab, Bitbucket and deployments on Forge, Cloud, Vercel without leaving your IDE.
dk CLI (Pro) — The fastest way to jump between projects from the terminal.

Configuration lives in a devkeepr.yml file you commit alongside your project, so your whole team gets the same hibernate/wake hooks for free.

Copied!

autoWake: true
syncUpstream: true
hibernateSchedule: "weekly"

hibernateCommands:
  - "rm -rf vendor"
  - "rm -rf node_modules"

wakeCommands:
  - "composer install"
  - "npm install"

Why we built it

Willem and I met for the first time at Laracon EU 2025. We'd been collaborating on NativePHP for months but never sat down together. Turns out we'd both been quietly building the same little desktop apps to scratch the same itches: a CI workflow monitor on his side, a deployment tracker on mine, both of us with our own scripts to clean up idle projects.

The conversation went from "we should share notes" to "we should just merge all of this into one app" in about ten minutes.

About a thousand commits later, here we are.

Devkeepr is also our excuse to push NativePHP to its limits. Everything we learn building it goes back upstream, so the framework gets better for everyone.

Try it

Devkeepr is free to use, with a Pro tier for the workflow monitor, deployment watcher and dk CLI.

Grab it at devkeepr.app.

I made a CLI tool to keep an eye on your dependencies: Whatsdiff v2 is out!

2025-10-29T00:00:00+00:00

I made this tool for myself. After a year of using it, I decided it needed some attention. v2 is here with a website and documentation.

What's Whatsdiff?
What It Does
How to you use it?
The real cool stuff
- Terminal UI
- Model Context Protocol (MCP) Server
Conclusion

What's Whatsdiff?

Ever run npm update or composer update and wonder what actually changed? Sure, you see package names scroll by, but what releases did you skip? What features or fixes did you get?

Whatsdiff answers that question. It's a CLI tool that compares your dependency files between commits and shows you exactly what changed—with release counts, version ranges, and even full changelogs.

What It Does

At its core, Whatsdiff analyzes composer.lock and package-lock.json files between git commits. Run it after an update, and you'll see:

Added, removed, updated, and downgraded packages
Multiple output formats: text, JSON, markdown, or an interactive Terminal UI
Full release notes for any package and version range
AI integration via an MCP server for assistants like Claude Code

How to you use it?

Basic usage is to just run it after an update of your dependencies.

Copied!

# Last update
whatsdiff

# Between specific commits, tags, or branches
whatsdiff between v1.0.0 v2.0.0

Want to see what's in those 5 Laravel releases?
It pulls from multiple sources: local vendor directories, GitHub Releases, or repository changelogs.

Copied!

whatsdiff changelog laravel/framework 12.33.0...12.36.1 --type=composer

You can also integrate this into CI/CD by using the check command to detect if specific packages have changed, exit codes make it perfect for automation (0 = true, 1 = false, 2 = error).

Copied!

# Check if a critical package was updated
if whatsdiff check livewire/livewire --is-updated --quiet; then
  echo "Running extended test suite..."
fi

You can also use it to create your own tools, use the JSON output format to create your own tool:

Copied!

whatsdiff --format=json

The real cool stuff

Terminal UI

Not a fan of scrolling through terminal output? Launch the TUI:

Copied!

whatsdiff tui

Browse packages, view details, and read changelogs with arrow keys and vim-style navigation.

Model Context Protocol (MCP) Server

For AI-powered workflows, Whatsdiff includes an MCP server that lets AI assistants:

Find compatible versions for dependency constraints
Fetch release notes between versions
Check available upgrades (patch, minor, major)
Analyze dependency trees

Perfect for integrating with Claude Code or other AI tools.

This helped Claude Code a lot when I need to upgrade a major version of a framework or a package.

Conclusion

If you want to give it a try, install it globally via Composer:

Copied!

composer global require whatsdiff/whatsdiff

I used it a whole year before talking about it. It's been super helpful to keep track of what changed in my projects.

When I take on a old project, the first thing I do is update dependencies and run whatsdiff to see what changed.

Copied!

npm update && composer update -W && npm run build && whatsdiff

Check it out at whatsdiff.app or github.com/whatsdiff/whatsdiff.

Your small VPS is almost full, clean up your journal logs

2025-10-21T00:00:00+00:00

Running a small VPS for your side projects? You might notice your disk filling up mysteriously. Before upgrading your plan or deleting important files, check your systemd journal logs – they're often the silent disk space killer.

The problem
Immediate fix
Permanent fix – Limit current log growth

The problem

By default, systemd journal logs everything: every service start, every error, every SSH attempt. On a small VPS, these logs can grow to several gigabytes without you noticing.

Check your current journal size:

Copied!

sudo journalctl --disk-usage

If you see anything over 1GB on a small VPS, you're wasting precious disk space.

In my case, /var/log/journal was eating 3.1GB on a 25GB VPS. That's 12% of my total disk space!

Immediate fix

Clean up old logs by deleting logs older than a certain time or limiting the total size:

Copied!

sudo journalctl --vacuum-time=7d # keep only last 7 days
sudo journalctl --vacuum-size=500M # keep only 500MB of logs

Permanent fix – Limit current log growth

To prevent this from happening again, you can configure systemd to limit the size and retention of journal logs.

Edit the journal configuration:

Copied!

sudo nano /etc/systemd/journald.conf

Uncomment and adjust these values to your preference:

Copied!

SystemMaxUse=500M
SystemKeepFree=1G
MaxRetentionSec=7day

Restart the service:

Copied!

sudo systemctl restart systemd-journald

That's it!

You just freed several gigabytes and prevented the problem from happening again.

Laravel Nightwatch: Log outgoing requests from Saloon

2025-06-17T00:00:00+00:00

Laravel Nightwatch has been released this morning, and on Unolia, we are using Saloon to make outgoing HTTP requests to our API. I wanted to log these requests in Nightwatch so I can see them in the dashboard.

UPDATE 21 June 2025: I pushed a PR to Saloon to add support for Laravel Nightwatch, which is available in Saloon v3.6.0 and later. Just use the following command and skip this article:

Copied!

composer require saloonphp/laravel-plugin "^3.6.0"

To log outgoing requests from Saloon in Laravel Nightwatch, you'll need to create a custom HTTP sender that integrates Nightwatch's middleware.

Following the Saloon and Nightwatch documentation, create this new class wherever you need it, for example, in app/Http/HttpSender/LaravelSender.php:

Copied!

<?php

namespace App\Http\HttpSender;

use Laravel\Nightwatch\Facades\Nightwatch;
use Saloon\HttpSender\HttpSender;

class LaravelSender extends HttpSender
{
    public function __construct()
    {
        parent::__construct();

        $this->handlerStack->push(
            Nightwatch::guzzleMiddleware()
        );
    }
}

And edit your config/saloon.php config file and change this line:

Copied!

'default_sender' => \App\Http\HttpSender\LaravelSender::class,

Simple!

This extends the Laravel Saloon HTTP sender to include the Laravel Nightwatch middleware.

Now, whenever you make an outgoing request using Saloon, it will be logged in Laravel Nightwatch.

Get the last login date of a user in Laravel

2025-06-13T00:00:00+00:00

By default, in Laravel, the last_login_at field is not available in the users table. I personally needed this information to run some background tasks.

The approach I chose is to use the Login event provided by Laravel Fortify.

For that, just create a new listener that will update the last_logged_at field in the users table whenever a user logs in.

The code
Some notes

The code

You can create a listener using the Artisan command:

Copied!

php artisan make:listener UpdateUserLastLoggedAt --event=Illuminate\Auth\Events\Login

And then implement the listener as follows:

Copied!

<?php

namespace App\Listeners;

use App\Models\User; 
use Illuminate\Auth\Events\Login;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Queue\InteractsWithQueue;

class UpdateUserLastLoggedAt implements ShouldQueue
{
    use InteractsWithQueue;

    public function __construct() {}

    public function handle(Login $event): void
    {
        /** @var User $user */
        $user = $event->user;

        $user->update([
            'last_logged_at' => now(),
        ]);
    }
}

By default, Laravel will automatically find and register your event listeners by scanning your application's Listeners directory. When Laravel finds any listener class method that begins with handle or __invoke, Laravel will register those methods as event listeners for the event that is type-hinted in the method's signature.

— https://laravel.com/docs/12.x/events#defining-listeners

Oh and don't forget to add the last_logged_at field to your users table migration:

Copied!

php artisan make:migration add_last_logged_at_to_users_table

Then, in the migration file, you can add the new column:

Copied!

Schema::table('users', function (Blueprint $table) {
    $table->timestamp('last_logged_at')->nullable();
});

And run the migration:

Copied!

php artisan migrate

Some notes

The last_logged_at field is nullable, so it will be null for users who have never logged in.
The UpdateUserLastLoggedAt listener implements the ShouldQueue interface, which means it will be queued for processing. This is preferable if you want to avoid slowing down the login process.
This works with Breeze, Jetstream, and Sanctum as they all use the Illuminate\Auth\Events\Login event.
If you wanna go further, the Login event also provides the following properties:
- string $guard – The authentication guard name.
- Authenticatable $user – The authenticated user.
- bool $remember – Indicates if the user should be "remembered".

How to automatically control the Raspberry Pi fan with Home Assistant

2025-06-09T00:00:00+00:00

I've set up a Raspberry Pi 4 with Home Assistant in my van. To manage its cooling, I want to programmatically control the fan through Home Assistant, triggering it only when the CPU temperature exceeds a specific threshold.

Hardware Setup
Software Requirements
Configuration
Result

Hardware Setup

I used a 5V fan connected to the Raspberry Pi's GPIO pins.

The little fan has three cables:

Red: Power (5V)
Black: Ground (GND)
Blue: Control signal

I connected the blue cable to GPIO 14, which is not a PWM-capable pin on the Raspberry Pi, but it works fine for simple on/off control.

Software Requirements

To control the GPIOs from Home Assistant, you need to install the following component in this order:

HACS - the Home Assistant Community Store
HA RPi GPIO - a plugin to control the GPIO pins of the Raspberry Pi

I let you read the documentation of these components to install them correctly.

Once you installed HACS, you can install the RPi GPIO plugin from the HACS dashboard on your Home Assistant instance.

Configuration

In your configuration.yaml file, add the following lines to configure the GPIO pin for the fan:

Copied!

switch:
  - platform: rpi_gpio
    switches:
      - port: 14
        name: "RPI Cooling Fan"
        unique_id: "rpi_cooling_fan"
        
climate:
  - platform: generic_thermostat
    name: RPI Cooling Fan Controller
    unique_id: rpi_cooling_fan_controller
    heater: switch.rpi_cooling_fan
    target_sensor: sensor.system_monitor_processor_temperature
    min_temp: 45
    max_temp: 80
    ac_mode: true
    target_temp: 55
    cold_tolerance: 0.5
    hot_tolerance: 0.5
    min_cycle_duration:
      seconds: 60 # Minimum time the fan should run before turning off
    keep_alive:
      minutes: 5
    initial_hvac_mode: "cool"

This configuration does the following:

Defines a switch for the fan connected to GPIO 14.
Sets up a generic thermostat that uses the fan as a heater.
Monitors the CPU temperature using the sensor.system_monitor_processor_temperature sensor.
Sets the target temperature to 55°C, with a minimum and maximum temperature range of 45°C and 80°C, respectively.
Configures the thermostat to run the fan for at least 60 seconds before turning it off, and keeps it alive for 5 minutes after the last activation.

Result

How I wasted 3 days upgrading to FluxUI

2025-03-26T00:00:00+00:00

I'm sorry for the clickbait title. I actually have only good things to say. I just wanted to catch people commenting before reading the article.

It started as a simple PR: add FluxUI to unolia.com and upgrade to TailwindCSS v4, and it ended as a big refactor of my project. Let me tell you my journey.

I started using FluxUI in other smaller projects, and I got used to it, especially for forms. It felt much cleaner and faster to work with Flux than anything else I had used before.

Before the migration
The big refactor
What’s next ?
Before and after screenshots
Conclusion
TLDR

Before the migration

Unolia.com is based on Jetstream, so most of the input was using <x-input> syntax shipped with Jetstream. Jetstream has been very convenient for jumpstarting the project, but over time I needed more.

I bought TailwindPlus several years ago, and it helped me a lot in making beautiful UIs. The thing is, if you don’t use components, you end up with big, unreadable HTML and inconsistency throughout your app. And if you want different input text with suffixes or prefixes, that means writing a lot of conditional code or duplicating code.

Anyway, for my case, I didn’t bother that much until now.

FluxUI is perfect for me. I use Livewire a lot, and it’s so tightly integrated. Design-wise, it is not far from TailwindPlus, although I preferred the dropdown design over it and other small details.

The big refactor

I began the conversion gradually, starting with a few input replacements.

Next, I worked on the dropdowns.

Then I noticed in the documentation that I could support dark mode.

After that, I lost track of what I was doing.

I became fully immersed in rewriting everything to enable dark mode support.

That meant rewriting some Jetstream components, reworking some layouts, some wording, well every detail of my app.

And the result is beautiful.

I had some head-scratching moments, especially with the way Jetstream handles its modal and also where Flux doesn't work outside Livewire sometimes.

I'm grateful to have made time for my robust test suite, especially the end-to-end tests.

Also, going through all my codebase and retesting my app allowed me to fix a couple of minor bugs or simply improve the experience.

What’s next ?

Now, I want to rewrite everything myself.

I never liked that Jetstream was a library instead of a starter kit. It's cool at the beggining but then it becomes difficult to personalize.

I realized that I added those dependencies to kickstart my project for the UI they provide, rather than for their features. I wanted Jetstream for team functionality and Filmant for notifications.

The same applies to wire-elements/spotlight. I never really used it, which is why I never purchased the pro version either. I had considered it, but after reading the documentation, I realized it wouldn’t fully offer the features I wanted. I believe a command bar needs to be truely well-made to be useful, which is why Raycast is so popular.

Anyway, I think FluxUI will allow me to create a nice command bar myself.

Also, some of those packages had versions with strict Livewire version locking (looking at you, Filament), and I didn’t like that because for a while a fix I needed was blocked because of that.

So I’ll continue removing dependencies from my projects

laravel/jetstream => Integrate the logic directely in my app.
wire-elements/spotlight => FluxUI Command
filament/notification => I will probably write my own package that leverages FluxUI (or not), but I already have specific needs that I'm sure are worth a PR.

All of this can obviously wait.

This is the kind of thing that does not even make it to the TODO list.

I just dive deep into when I feel it.

Before and after screenshots

From the dashboard, you can see how the header and the navigation are much more balanced than before.

I replaced the filament table with a FluxUI table. It's much easier to personalize.

For the domain: more space, cleaner tabs, button weights revised, record type aligned to the right. Also using the new avatar component for the favicon of the website (future small feature).

Again, small details, but they help see it clearer. Also, it took me some time to add a dark mode to my custom radio buttons.

Conclusion

So this is how a simple "let’s add flux" turned into "let’s refactor my entire project for no reason."

Why this title, though? Because using 3 days of my free time to add 0 value to my customers is arguably a waste of time.

But I don’t regret a single minute.

It was great. unolia.com has a clearer UI, a better UX, and most importantly, cleaner code.

I can now focus back on other features.

To conclude this article and if you like the TALL stack:

100% recommend FluxUI.

TLDR

Refactoring to Flux reduced the number of lines of code in my project while giving me more features.
unolia.com now supports dark mode.
Jetstream is about to disappear completely from my project. I embrace the new starter kit way.
Tests are mandatory before rewriting your app. However, tests can’t see everything. Manual tests are still mandatory in some cases.
Refactoring also means going through all your app and improving every detail of it

Post scriptum:

I'm wondering, should I do a dark mode here?

Nah, I think I have better things to do.

Host your own AI Server using Proxmox and Ollama and connect PHPStorm to it

2025-02-13T00:00:00+00:00

This is a guide on how to set up an AI server on Proxmox, connect it to PHPStorm, and use Tailscale for remote access.

I have a rather beefy server at home that I use mostly as my Windows gaming machine. It runs Proxmox because, from the beginning, I knew I wanted a server where I could run VMs and containers from a distance.

You may not know, but I travel a lot with my van, and I can't bring all my equipment with me.

For context, here are the hardware specs of my server:

CPU: AMD Ryzen 9 5900X 12-Core Processor
RAM: 32GB (I need to upgrade it to 128GB or something)
GPU: Nvidia RTX 3070 + Nvidia RTX 3060 Ti
Storage: 2TB NVMe + 2x 60GB SSD

Create a new VM
Install required Nvidia CUDA Drivers and Toolkit
Docker Installation
- Install Nvidia Container Toolkit
Use Ollama for running AI models
- Basic Usage of Ollama
- Make Ollama listen outside of localhost
OpenWebUI Installation
Tailscale
PHPStorm Integration
Update Ollama and OpenWebUI
Conclusion

Create a new VM

We begin by setting up an Ubuntu VM on Proxmox, allocating the necessary resources for your AI server. You can download an ISO and create a new VM from it with the console, but I prefer to use a script to automate the process.

The script comes from the community-scripts repository, and it's a simple bash script that you can run on your Proxmox server.

Run the following script to deploy the Ubuntu VM:

Copied!

bash -c "$(wget -qLO - [Proxmox script](https://github.com/community-scripts/ProxmoxVE/raw/main/vm/ubuntu2404-vm.sh))"

The script will ask you for some information; here is what I used:

I allocated 512 Go storage because LLVM models are huge.
I gave it 24 576 MB RAM. I tried with 32 GB, but it made my server crash. Keep some RAM for the host. Ollama is great at distributing the load on the CPU and GPUs. This gives me plenty of room to test large models.
Do not start the VM immediately after installation.

After the installation is finished, in the Cloud-Init tab:

Configure the user user, set a password, and/or add your public SSH key.
I chose to use DHCP for the network configuration because I will only access it from my Tailscale network.
Click on Regenerate image

Before starting the VM, go to the Hardware tab and add your GPU to the VM.

Add -> PCI Device -> Raw Device -> Your GPU
Check All functions
In the advanced options, check ROM-Bar
Start the VM.

After that, you have a fresh Ubuntu VM ready to be used.

I took some time to install ZSH and Oh-My-Zsh, but do as you wish.

Install required Nvidia CUDA Drivers and Toolkit

Now it's time to configure this server to work well with AI models.

Start by confirming that your GPU is detected with:

Copied!

lspci | grep -i nvidia

Now we need to install Nvidia's CUDA Toolkit and drivers.
The instructions for your machine can be found here CUDA downloads page

For our Ubuntu 24.04, we will use the following commands:

Copied!

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update
sudo apt-get -y install cuda-toolkit-12-8
sudo apt-get install -y cuda-drivers

Now we need to reboot the server to apply the changes:

Copied!

sudo reboot

Set the environment variables:

Copied!

export PATH=/usr/local/cuda-12/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

And verify the installation:

Copied!

nvcc --version
nvidia-smi

That last command shows the GPU usage and the CUDA version.

Later you can watch it live with:

Copied!

watch -n 0.5 nvidia-smi

There is also a command called nvtop that you can install with:

Copied!

sudo apt-get install nvtop

Docker Installation

For some software that we will use, we need to install Docker.

Follow the steps detailed in the Docker installation guide:

or copy and paste the following commands:

Copied!

# Add Docker's official GPG key:
sudo apt-get update
sudo apt-get install ca-certificates curl
sudo install -m 0755 -d /etc/apt/keyrings
sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
sudo chmod a+r /etc/apt/keyrings/docker.asc

# Add the repository to Apt sources:
echo \
"deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \
$(. /etc/os-release && echo "${UBUNTU_CODENAME:-$VERSION_CODENAME}") stable" | \
sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt-get update

# Install Docker
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

Confirm the installation by running:

Copied!

sudo docker run hello-world

Install Nvidia Container Toolkit

I don't use Ollama in a Docker container, but you may want to. Or maybe you need to run some other containers that need access to the GPU. This section ensures that you can use the GPU in a container in the future.

Nvidia container toolkit documentation

Follow the instructions for your distribution.

Use Ollama for running AI models

Ollama is a powerful tool for running AI models on your server. It provides a simple command-line interface to run various AI models with ease.

It also comes with an API that can be used by other applications.

One feature that I like is that it can distribute the load on multiple GPUs and CPUs. I tried to run models that only use GPU; the rule of thumb is to use models that can fit in the combined memory of all GPUs.

This is our true AI server.

I decided to install Ollama directly on the VM, but you can also use a Docker container.

Install Ollama with the provided script:

Copied!

curl -fsSL https://ollama.com/install.sh | sh

Basic Usage of Ollama

Run various AI models with commands such as:

Copied!

ollama run llama3.2 --verbose
ollama run deepseek-r1:14b --verbose
ollama run deepseek-r1:8b --verbose

I like to use the --verbose flag to see stats like tokens per second and samples per second after every answer.

You can also just pull the models without running them:

Copied!

ollama pull mistral

You can also see how the load is distributed on the GPUs and CPUs with:

Copied!

ollama ps

Make Ollama listen outside of localhost

This step will be useful if we want our PHPStorm to connect to Ollama.

Copied!

systemctl edit ollama.service

Add the following lines at the beginning of the file:

Copied!

[Service]
Environment="OLLAMA_HOST=0.0.0.0:11434"

OpenWebUI Installation

OpenWebUI is a web interface that gives you a ChatGPT interface to interact with your AI server.

OpenWebUI GitHub repository for detailed installation instructions.

Here is a quick guide to install OpenWebUI:

Copied!

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

You can now access OpenWebUI at http://your-server-ip:3000.

The first account created will be the admin account.

Tailscale

Tailscale is your personal VPN that makes it easy to connect your devices through internet without any port forwarding.

To install Tailscale, the easiest way is to go to your Tailscale admin console and get the command to install it on your server.

Then the next step is to tell Tailscale to serve OpenWebUI:

Copied!

sudo tailscale serve --bg 3000

You can now access OpenWebUI from anywhere with your Tailscale Magic DNS with HTTPS.

Example: https://aiserver.tail1234567.ts.net

PHPStorm Integration

For PHPStorm to connect to Ollama, we need to install the offical PHPStorm AI plugin.

In the configuration, you can add your Ollama server with the IP and port.

Update Ollama and OpenWebUI

To update OpenWebUI automatically, you can use Watchtower.

Here is a command to update OpenWebUI once:

Copied!

sudo docker run --rm --volume /var/run/docker.sock:/var/run/docker.sock containrrr/watchtower --run-once open-webui

And here is a command to update OpenWebUI every hour:

Copied!

sudo docker run -d \
--restart always \
--volume /var/run/docker.sock:/var/run/docker.sock \
containrrr/watchtower --interval 3600 open-webui

The only way to update Ollama is to run the install script again.

Here is a command to update Ollama every week:

Copied!

sudo crontab -e

Add the following line to the file:

Copied!

0 0 * * 0 curl -fsSL https://ollama.com/install.sh | sh

Conclusion

And voilà!

You now have a personal AI server that you can access from anywhere with PHPStorm and OpenWebUI.

You should dig into the documentation of Ollama and OpenWebUI to see all the features they offer.

Examples of advanced features you can explore:

Image generation
Indexing documents with embeddings
Chat with ChatGPT and local models at the same time
Use multi modal models like llama3.2-vision

2024 : Year in review

2024-12-27T00:00:00+00:00

I definitely have not followed all my wishes from last year, but who does? This year was my 30th year on this big rock, and it marked a turn, more than I thought it would.

Traveled a lot
Started my own company
Wrote a lot more
I got payed for Open Source work
Now I'm poor
But somehow happy
What's next then?

Traveled a lot

If you don't know, I'm living in a van.

I converted it myself and traveled through Western Europe.

But not much this year because I had some paperwork to do, and it took me 9 months to get through.

I still traveled a lot, though:

January started strong: one awesome ski trip with my friends ⛷️
In February, I attended Laracon EU in Amsterdam!
In February, I also spent a week in Copenhagen with my friends.
March: another ski trip with my cousins! ⛷️ I'm addicted!
May and June: Festival 🎶 !
In July, we had the Olympics in Paris. It was fun to watch, follow, and attend too. Really good vibes. Surprised.
In August, I threw a big party for my 30th birthday!
In October, I spent two weeks in Turkey.
In November, I traveled in my van in the south of France.

I thought I didn't do much this year, but looking back at my photos:
I had a pretty good year of travels.

Also, that's the trick for your yearly review:
Take your phone and look back at your photos.

Started my own company

I started a SaaS. Every developer's dream, right?

Well, this year, I realized mine. I launched the first version of unolia.com.

After being at Laracon EU, I made the decision to finally start working on my big idea. I wanted a proof of concept live as fast as possible, so the first three months were very productive. I limited my product to the basics I would need: connecting every major domain provider.

And in June, an open beta was online!

This is not even 1/5 of the ideas I had in mind. I knew I would need to break the project down into reachable goals, so I started with domains. But the ultimate goal is to connect every service you have and write automation between them.

Then... I don't know, it slowed down. I worked on it every month, but not enough.

In October, I took a week to write actual marketing content.

And in November, I integrated Stripe, and it's now a paid product.

I made myself a promise to work at least 1 complete year on it to give it a chance.

I think now that it needs 2 years because of what I see from other similar stories and also because I don't think I worked enough on it this year.

Coding a SaaS is not hard, creating a design can be fun, doing marketing is interesting, and even creating a legal structure is okay.

But doing it all at the same time... boy oh boy, it's exhausting.

Wrote a lot more

Last year, I published 8 articles here, well 9 with this one.

I want to double that this year.

I love writing. I write a lot in my bed at the end of the day or on a public bench after long hours of driving. Even writing documentation for my code and projects, I just love it.

I secretly hope to write a book someday.

I feel like I'm better at writing in English, and I'm also better at writing in French. I write my adventures in my notes, and sometimes I even make rhymes for the fun of it.

I also tweeted more.

This led me to be more visible on social networks and slowly built an audience.

I got payed for Open Source work

Last year, I published 4 open-source projects. This year, 8 more!

I like sharing, writing a good API, and creating beautiful READMEs. If I have a coding problem, I'll create a solution for it. It's a good exercise, and it's gratifying because when it's done, it's done (unlike big projects).

Yes, I know that after that, there is some maintenance and updates involved, but so far, I'm enjoying that too.

I don't make any money off it. It was never the goal.

BUT I still earned some money!

Not on my own project, but on NativePHP. Simon reached out to me and wanted to thank me for some of my PRs on the nativephp.com website.

I accepted, and shortly after, I had 250€ in my account.

I was honored and surprised.

It made me so happy! I told all my friends, even though half of my friends didn't understand what open source is.

I should have cried out loud on Twitter too. But I wasn't sure if I should. I hadn't been in this situation before. A "stranger" giving me money for something I happily did for free. Still today, I don't know why I kept it silent.

Because I think it's a movement we have to support. Open-source shouldn't mean free. We're helping each other. It's only fair to help each other financially too.

And everyone should know how awesome Simon is. He's just a good guy. I believe in him, and I hope that he will succeed in every project he decides to go after next year.

I wanna do more open-source, I'm good at it.

Now I'm poor

Talking about Money, I never been so low.

All my savings are gone.

It's been almost 3 years since I've been unemployed.
I took 2 years to travel and 1 year to discover who I am and what I want.

Quite a long pause you may say but for the outcome, it was worth it.
(you'll see next chapter)

In November: I had barely enough money to drive back to my hometown.
Mid December: Don't have enough to pays my insurances.

I should be at the bottom of a pit being eaten by anxiety.

In fact, I almost fell into it.

But I did see it coming one evening. From my bed, I woke up again. And made a list of all my expenses and earnings. Gone through all my accounts, savings, everything.

The result was.. concerning to say the least.

But at least I stopped burying my head in the sand and I had a short-term plan.

So I sold some unused hardware.
And later, all my remaining cryptocurrencies.
I won a little amount of money here and there.

By my estimations, I can live until mid-February. Then... well, I don't know.

I started searching for short missions and peeked at some real jobs, but I don't want to go back to any company. I can't handle having a boss anymore. I want partners. I only want to work with people I could eat lunch with every day (in the sense that I like them).

Having no money has been "good."

Living in a van taught me how valuable water, hot water, electricity, and comfort are.

Being poor taught me how to live with a minimum amount. With my current way of life, I could work on open source all day, every day, for just about ~2000 € per month (before taxes 😅).

Of course, I want more. Just look back at every trip I've taken last year.

But at least I know how to live with less now.

This is now my new minimum I have to achieve.

But somehow happy

The thing is, I think I finally found how I want to live this life.

Far less alcohol (and drugs)
I used to say that alcohol has no effect on me. And yes, I can drink a considerable amount before losing control. But I also discovered that it was one of the culprits for my degraded health. I don't want to be drunk anymore, even for a good party or a big festival. Life is short; I don't want to waste any days by being hungover. I feel like I'm too old for this. It was fun, but now my body is taking too long to recover.

Live outside
Ok, obviously I don't want to live in a van for the rest of my life. But I like being in contact with nature. If I can, I would continue to travel, go backpacking, and cycle. But the ultimate goal is to have enough money to buy land, have a tiny house, and a garden. Programming with a view.

Focus on myself
I'm done doing things for the love of others. I don't want to be defined by others. I don't want to please anyone. It seems extreme said like this, but I hurt myself a lot just to find someone to love me back. I have made peace with myself. I'm okay being alone most of the time.

Doing what I love
I love tinkering. That's what makes me happy. Domotizing a house, upgrading a car, converting a van, and making web, CLI, and desktop apps are some of the things I have made in the last 3 years, and I absolutely loved doing them. When I was younger, I said that I wanted to be an inventor. And look mom, I did become one!

What's next then?

Last year I had money but I wasn't really happy.
This year I'm happy but I don't have any money.

Funny how the life works, huh.

So.. Make some money.

Yep, that's the goal..

I still strongly believe in unolia.com. I have many things I would like to solve with it and It didn't get the time it deserved.

You'll see me at Laracon EU.

Last year I did go back from Amsterdam with the motivation to start unolia.com. This year will mark the 1st anniversary of me starting the project and I would like to go out from there with the motivation to double down on it.

I also strongly believe in NativePHP. I think I have submitted at least one PR every week for the last two months. I have ideas for apps to sell and ideas for advanced packages to sell. It has a bright future; I can guarantee you that. I would be involved in this project as much as I can.

But the thing is that I have to start earning money fast.

Like, really fast.

I'm finally good with myself, good with my mind, and my body.

So having a good financial situation is my way to an happy stable life.

I "only" have to reach the 2000 € / month revenue.

I'll make it, for sure.

One way or another, I don't have a choice anymore.

But I think I will stop open-source for a bit and make real money.

Sadly there isn't enough Simon out there to make open-source worth it.

I don't want to ask, but if you read this far and want to support my efforts, I would be grateful! Depending of what rewards you want:

You can sponsor me through Github Sponsors
Or subscribe to unolia.com

Either way, it would be life saving right now.

And please, comment below.

Even just a smiley would mean a lot to me, it shows real people are reading it.

I'm going to end this article by wishing you all the best.

If you are that far into listening to a man struggle, you're one of the best people this planet has to offer.

I hope you’re okay and that you take care of yourself.

I wish all the you success and happiness.

See you in 2025.

Post scriptum.
I read somewhere that the decisions taken in the European Commission were better because most of them weren't using their main language. The experiments showed that communicating in another language put distance between the writer and the text, eliminating emotions in the decision-making.

At some point in this article, I asked myself why I don't ever express myself that much outside of my blog. I think writing in English may help me have the courage to write some of those lines.

So I'm grateful for this. Hope you liked it.

My take on Bluesky

2024-12-04T00:00:00+00:00

Well, here we go again. This is supposed to be a technical blog, but I enjoy giving my opinion more and more. This is a direct reply/reaction to the discussion Ian and Aaron had on Mostly Technical. An alternative title is "Twitter vs. Bluesky." You've been warned.

Mainstream
Discover VS Following feed
Sharing a link
→ TLDR

Mainstream

What I consider mainstream is that my IRL friends are on it.

Bluesky is not mainstream yet.

And I hope it will never be.

When I registered on Twitter in 2009 (that long 😲), there were only tech people on it. A few years later, I remember my IRL friends saying, "Did you see this new thing called Twitter?" and my reply was always, "Yeah man, I've been on it from the beginning, years ago, and it's great. "

Some years later, everyone is telling me that the content is shit and they don't go anymore. But I'm still happy there because of one thing: I handpicked every people I follow.

I hear Aaron complaining about politics and sports stuff on Twitter/Bluesky, and Ian about some gory things that pop up. I never had those. I mean here and there, but not as much as you seem to have in the American feed algorithm.

Maybe the problem comes from who you are following. I remember following a friend who tweeted about the NBA and football (⚽️ the real one with a ball at your foot) all day long, and it was poisoning my beautiful tech Twitter. I unfollowed him immediately and added football as a muted word ^^

Muted words are awesome.

Mainstream people are noisy.

Discover VS Following feed

I always preferred the Following feed.

I want my feed to be chronological so I can see the content from accounts I chose to follow. The same goes for YouTube btw.

I discover new people through retweets. Much like real life, you meet someone new through a common party or recommendation. It the retweeted tweet is cool: you go to their profile, and if their content seems to be of interest to you, you click on that "Follow" button. Easy.

And you know what the cool part is? You can "Unfollow" when it's no longer in your interest.

I do click on the "Discover" and "For You" tabs from time to time when I finish reading my Following tab, but only when I'm bored.

We're not meant to read every f*cking thing that happens in the internet.

Just focus on what you enjoy reading and learning.

Sharing a link

I mentioned how great the early Twitter era was, but there’s something I didn’t tell you: it replaced my RSS feed. Every major blog had a bot account that posted new articles, and I loved scrolling to the top to read my news.

However, someone at Twitter decided that this was no longer a good thing..

I think that lowering the visibility of tweets with links was a terrible mistake. I know why they did that, still a mistake. I hated every moment I had to click on a tweet to see the link attached to it.

Also, I believe that displaying views and monetising them significantly contributed to making Twitter a worse place. Tweeting should not be a job or a daily goal.

Seeing liked posts on my feed made it worse too. If the person wanted me to know, he would have clicked on the retweet button.

I don't know what could make Twitter great again.

And it's too early to say if Bluesky is that shiny.

→ TLDR

In 2009, Twitter replaced IRC and RSS for me.
Bluesky is great because of the lack of algorithms.
Stay in the Following tab, read it as your newspaper, and stop doom scrolling.
Muted words are important on every social network.
You can control what you see on social networks.
Stats are what make people chase vanity.

Still don't know who you are and why you read such articles. Hope you enjoyed!

Bonus, thanks to Bluesky comment Statamic addon, you can comment on my blog now.

My take on Statamic's future

2024-10-18T00:00:00+00:00

Recently, Jack McDade (creator of Statamic) started a discussion on the future of Statamic. I have been using Statamic for a while now and I have some thoughts on the future of Statamic.

You are currently on a Statamic website. Statamic is great, especially if you are a developer, even more if you are a Laravel developer.

Flat file is awesome
Antlers is cool, but...
Pricing and market needs
Statamic should definitively have cloud hosting
TL;DR

Flat file is awesome

I like writing my articles in Markdown. I use the control panel to create and manage my content, but I write the content of my articles in PHPStorm. I'm a developer, so writing my articles from my IDE feels like home. I have my shortcuts, my plugins, my themes, my settings, my everything. I even use GitHub Copilot when I don't remember a word.

That said, for the future, and to make it more accessible to non-developers, I think Statamic should run on SQLite by default.

Antlers is cool, but...

It had never been properly supported in my IDE, and it becomes unreadable and broken every time I try to auto-format it. Autocomplete will also help me learn the syntax. When I use a package or framework, there are a lot of functions I never read the docs for; it's just in my IDE, and I understand what it does by reading the function signature.

I have never tried to make a Statamic website with Blade, but I understand there are some limitations, and the documentation recommends sticking with Antlers, so that's what I do.

I'm in between; I think it's cool, but I'm kind of frustrated by it.

Pricing and market needs

I remember the first time I heard about Statamic. I did go to the website and I saw the price and said, "No way it can't be that worth". I judged it without even trying it.

And today, I think the price is totally fair!

I don't pay for Statamic (sorry, Jack 😬). I don't need the pro features except for Git automation, and I use a basic bash script with a cron job for that. I may have purchased a license if there had been a no-brainer $99 license that included Git automation and multiple users. Soon enough, I will purchase a license for an upcoming multilingual website I'm working on.

From what I see of the WordPress market in France, it's split like this:

Non-tech people using WordPress to create their websites because family and friends told them to use WordPress. They use a hosting service that provides a one-click install of WordPress.
Independent designers who create websites for small businesses. They use web hosting services that provide a one-click install of WordPress and customize it through themes, plugins, and FTP.
Web agencies that create websites for medium to large businesses. They have their own servers and customize WordPress beyond its original capabilities by connecting it to other services, creating custom themes, and plugins.

So my point of view is:

Free should be pushed more to the front page
$99 for Git automation and multiple users
Hosting service with monthly subscription

Statamic should definitively have cloud hosting

This is a no-brainer, there is a market for it. It's even been in my Trello of ideas to create a Statamic hosting service. Even a simple one to start that basically only does what Forge does for Laravel: a one-click install.

But I also think it should go a little further:

It should have its own starter kit to start a blog, a portfolio, a company website, etc.
It should be easy to install addons from the control panel (curated by Statamic)
Themes should be editable from the control panel for non-developers
It should support static caching and be enabled by default

TL;DR

I think Statamic should:

Run on SQLite by default
Improve Antlers support in IDE
$99 for Git automation and multiple users
Hosting service with monthly subscription

If Jack reads this or anyone else really: I will be happy to discuss or develop any of these points with you in person Laravel EU or on any of my social networks.

Upgrade your Filament notifications by live updating them

2024-09-10T00:00:00+00:00

While working on unolia.com I made a little upgrade to my notifications! I wanted my user to have feedback on the progress of the tasks. In one PHP class, no additional table, clean API.

Prerequisites
The problem
The solution
Security concern
How can we do more?

Prerequisites

Laravel 11 app (with at least job batches and anonymous events)
Laravel Reverb installed and configured
Filament v3 Notifications

The problem

To better understand what we are trying to improve, here is the current Livewire function that is called when the user requests issues to be checked.

Copied!

#[Renderless]
public function checkIssues()
{
  $records = $this->zone->records()->get();

  foreach ($records as $record) {
      dispatch(new CheckZoneRecord($record));
  }

  Notification::make()
    ->title('New issues checks have been requested')
    ->success()
    ->send();
}

This function does two things:

Sending CheckZoneRecord jobs to the queues
Sending a visual notification, more like a toast, telling the user that the checks have started.

Depending on the part of my application, they also receive a notification when the background task is done. But here I have a batch of jobs and there is no way that one job will know and be responsible to send this success notification.

That was the first challenge: how to tell the user that checks are done? Easy. Use job batch to send a notification when it's done.

Copied!

Bus::batch($jobs)->finally(fn() => sendSuccessNotification())

But like I said, I already had this kind of user experience in my app.

I wanted more!

I wanted a progress bar!

The solution

I wanted the solution to be as seamless and transparent as possible, but in the end, I had to slightly customize Filament.

Customize Filament

We are going to use our own view for the notifications, so go to AppServiceProvider.php and thoses lines int the boot() function:

Copied!

Notification::configureUsing(function (Notification $notification): void { 
  $notification->view('components.notification');
});

Original file:
vendor/filament/notifications/resources/views/notification.blade.php

Now create the component with

Copied!

php artisan make:component notification --view

Copy and paste the content of the original file and and modify the body part:

Copied!

@php 
    use Filament\Notifications\Livewire\Notifications;
    use Filament\Support\Enums\Alignment;
    use Filament\Support\Enums\VerticalAlignment;
    use Illuminate\Support\Arr;

    $color = $getColor() ?? 'gray';
    $isInline = $isInline();
    $status = $getStatus();
    $title = $getTitle();
    $hasTitle = filled($title);
    $date = $getDate();
    $hasDate = filled($date);
    $body = $getBody();
    $hasBody = filled($body);
@endphp

<x-filament-notifications::notification 
    :notification="$notification"
    :x-transition:enter-start="
        Arr::toCssClasses([
            'opacity-0',
            ($this instanceof Notifications)
            ? match (static::$alignment) {
                Alignment::Start, Alignment::Left => '-translate-x-12',
                Alignment::End, Alignment::Right => 'translate-x-12',
                Alignment::Center => match (static::$verticalAlignment) {
                    VerticalAlignment::Start => '-translate-y-12',
                    VerticalAlignment::End => 'translate-y-12',
                    default => null,
                },
                default => null,
            }
            : null,
        ])
    "
    :x-transition:leave-end="
        Arr::toCssClasses([
            'opacity-0',
            'scale-95' => ! $isInline,
        ])
    "
    @class([
        'fi-no-notification w-full overflow-hidden transition duration-300',
        ...match ($isInline) {
            true => [
                'fi-inline',
            ],
            false => [
                'max-w-sm rounded-xl bg-white shadow-lg ring-1 dark:bg-gray-900',
                match ($color) {
                    'gray' => 'ring-gray-950/5 dark:ring-white/10',
                    default => 'fi-color-custom ring-custom-600/20 dark:ring-custom-400/30',
                },
                is_string($color) ? 'fi-color-' . $color : null,
                'fi-status-' . $status => $status,
            ],
        },
    ])
    @style([
        \Filament\Support\get_color_css_variables(
            $color,
            shades: [50, 400, 600],
            alias: 'notifications::notification',
        ) => ! ($isInline || $color === 'gray'),
    ])
>
    <div
        @class([
            'flex w-full gap-3 p-4',
            match ($color) {
                'gray' => null,
                default => 'bg-custom-50 dark:bg-custom-400/10',
            },
        ])
    >
        @if ($icon = $getIcon())
            <x-filament-notifications::icon
                :color="$getIconColor()"
                :icon="$icon"
                :size="$getIconSize()"
            />
        @endif

        <div class="mt-0.5 grid flex-1">
                  
            @if ($hasTitle)
                <x-filament-notifications::title>
                    {{ str($title)->sanitizeHtml()->toHtmlString() }}
                </x-filament-notifications::title>
            @endif

            @if ($hasDate)
                <x-filament-notifications::date @class(['mt-1' => $hasTitle])>
                    {{ $date }}
                </x-filament-notifications::date>
            @endif
@if ($hasBody)
    <x-filament-notifications::body
        @class(['mt-1' => $hasTitle || $hasDate])
    >
            <div
                x-data="{body: @js(str($body)->sanitizeHtml()->toString() )}"
                x-html="body"
                x-init="
                Echo.private('App.Models.User.{{ auth()->user()->id }}').listen('.UpdateNotificationBody', (event) => { if (event.id == '{{$notification->getId()}}') body = event.body });
                Echo.private('App.Models.User.{{ auth()->user()->id }}').listen('.CloseNotification', (event) => { if (event.id == '{{$notification->getId()}}') setTimeout(() => $dispatch('close-notification', {id: event.id}), 2000); });
                "
            >
                {{ str($body)->sanitizeHtml()->toHtmlString() }}
            </div>
    </x-filament-notifications::body>
@endif
            @if ($actions = $getActions())
                <x-filament-notifications::actions
                    :actions="$actions"
                    @class(['mt-3' => $hasTitle || $hasDate || $hasBody])
                />
            @endif
        </div>
                      

        <x-filament-notifications::close-button/>
    </div>
</x-filament-notifications::notification>

Here we basically added an alpine component that listent to two events:

UpdateNotificationBody for updating a notification
CloseNotification for closing the notification when progress == 100%

Create a new helper class 🎨

Put it wherever you want. If you read the code, it's self-explanatory.

I made the API simple enough so my users will experience the same notification throughout the project, but you can expand it if necessary. The end success notification is a bit more customizable.

Copied!

<?php

namespace App\Services\BackgroundTasks;

use App\Models\User;
use Closure;
use Exception;
use Filament\Notifications\Notification;
use Illuminate\Bus\Batch;
use Illuminate\Bus\PendingBatch;
use Illuminate\Contracts\Support\Htmlable;
use Illuminate\Support\Collection;
use Illuminate\Support\Facades\Broadcast;
use Illuminate\Support\Facades\Bus;

class BackgroundTasks
{
    protected Notification $notification;

    protected PendingBatch|Batch $batch;

    protected User $user;

    protected Notification $successNotification;

    protected array|Collection $jobs;

    public function __construct(array|Collection $jobs, User $user)
    {
        $this->jobs = $jobs;
        $this->user = $user;

        $this->notification = Notification::make()
            ->body('Starting...')
            ->icon('heroicon-o-cog-8-tooth')
            ->info()
            ->persistent();

        $this->successNotification = Notification::make()
            ->title('Task executed successfully')
            ->success();
    }

    public static function make(array|Collection $jobs, User $user): self
    {
        return new self($jobs, $user);
    }

    public function setTitle(string|Closure|null $title): self
    {
        $this->notification->title($title);

        return $this;
    }

    public function setIcon(string|Htmlable|Closure|null $icon): self
    {
        $this->notification->icon($icon);

        return $this;
    }

    public function dispatch(): self
    {
        $notificationId = $this->notification->getId();
        $userId = $this->user->id;
        $successNotification = $this->successNotification->toArray();

        $this->batch = Bus::batch($this->jobs)
            ->progress(function (Batch $batch) use ($notificationId, $userId) {
                self::updateNotificationProgress($batch, $notificationId, $userId);
            })
            ->finally(function (Batch $batch) use ($notificationId, $userId, $successNotification) {
                self::updateNotificationProgress($batch, $notificationId, $userId, $successNotification);
            })
            ->dispatch();

        $this->notification->send();

        return $this;
    }

    public function setSuccessNotification(Closure $callable): self
    {
        $this->successNotification = $callable($this->successNotification);

        return $this;
    }

    public function getBatchId(): string
    {
        if ($this->batch instanceof PendingBatch) {
            throw new Exception('Batch has not been dispatched yet');
        }

        return $this->batch->id;
    }

    public function getNotificationId(): string
    {
        return $this->notification->getId();
    }

    public static function updateNotificationProgress(Batch $batch, string $notificationId, int $userId, ?array $successNotification = null): void
    {
        $body = $batch->progress().'% done';

        Broadcast::private('App.Models.User.'.$userId)
            ->as('UpdateNotificationBody')
            ->with([
                'id' => $notificationId,
                'body' => $body,
            ])
            ->sendNow();

        if ($batch->finished()) {

            if ($successNotification) {
                Notification::fromArray($successNotification)->broadcast(User::find($userId));
            }
            Broadcast::private('App.Models.User.'.$userId)
                ->as('CloseNotification')
                ->with(['id' => $notificationId])
                ->send();
        }
    }
}

Use it on your project

Now we take our previous code, and use our knewly created BackgroundTasks helper!

Copied!

#[Renderless]
public function checkIssues()
{
  $jobs = $this->zone->records()->get()
    ->map(fn ($record) => new CheckZoneRecord($record));

  BackgroundTasks::make($jobs, $this->user)
    ->setTitle('Checking for new issues')
    ->setSuccessNotification(function (Notification $notification) {
		return $notification->title('Issues checked !');
    })
    ->dispatch();
}

And here it is !

Security concern

Since I know that I'm not using any HTML in my notifications, my version uses x-text. In this article, I changed it to x-html so it will not break your application. That should not be a problem security-wise, but be aware that the body sent through the websocket is not sanitized by str($body)->sanitizeHtml() neither by AlpineJS.

Basically, if your websocket server is compromised (or if you use a Public channel, I'm not sure): the attacker could send any HTML he wants in your DOM.

If you don't use any HTML in your notification, change it to x-text to be sure. Or read the last part of this article where I explain how to do even better by customizing even more Filament.

How can we do more?

I don't know what's on the roadmap of Filament v4 but I hope this feature will be included in the next release. Or at least I hope this article will push someone make a PR to Filament.

I almost made a package of it but I don't think this version is worth it.

If you are brave enough, here is a list of things I would do differently:

Support for database notifications: For now, only broadcasted notification support this feature but it should be easy enought to update the notification in the database.
Add a new variable to mark the notification as updateable so they could be rebroadcasted if you open a new tab. Like a really persistent notification.
Make a sub livewire component for each Notification to clean the code and have them listen to events dedicated to them. This will also help preventing XSS.

Hope you liked this article!

Review of my first time live streaming

2024-07-23T00:00:00+00:00

Today was my first time streaming live on internet. I wanted to livestream for a long long time. When I was gaming, when I was travelling and now while I'm coding.

Several weeks ago I started looking to see how difficult it would be to do a live coding session. And after an hour of tinkering I was already up and running.

Subject of the stream
Quick look at the setup
The good
The bad
The ugly
What's next ?

Subject of the stream

I need a little menubar app to start and stop my VM installed on a Proxmox server. Right now I'm using wakeonlan to start the server, wait several minutes, then I go to the web ui to start or stop each VM. Then after I'm done I shut down the server.

I like the idea to have a dedicated app and I don't want to wait for the server to boot to start the VM.

I want a user-friendly app I could upgrade with more features in the future.

Quick look at the setup

The hardware is pretty straight forward:

Computer: MacBook Pro 2020 (i7, 16GB, 512GB)
Camera: iPhone 13 mini
Microphone: AirPods Pro
External monitor to monitor the live and have the chat and other thing off-screen

For the software, I took the easy way:

Streaming software: OBS for capturing the screen and the camera.
Streaming service: Restream to stream to Twitch, YouTube and Twitter
OBS plugin: OBS Composite Blur

The good

OBS has come a long way! I followed the development by installing it from time to time, but I never really used it. It's easy to use and powerful. I was able to set up the stream in less than 20 minutes.

I used a plugin to blur the screen when I knew passwords or other sensitive information would be displayed. With OBS you can put a shortcut on basically everything.

For the next stream I want to repeat theses:

Using my iPhone as a camera was a good idea. The quality is good and I can put it loosely somewhere on the desk It follows me. You can apply filters to the camera feed directly in MacOS (blurry background, studio light, etc..).
Having a second monitor was really helpful.
I think that lowering my screen resolution helped a lot with the performance and the clarity of the stream.
Like expected, I was focused on the app the whole time of the stream. It was one of my reasons to start streaming. I can be distracted easily and I wanted to force myself to be focused on the task at hand.
Keep the minimum of apps open and turn on Focus mode on MacOS.

The bad

And here is a list of things that went wrong and that I need to improve

It can be awkward for the viewers to have nothing happening on the screen while I'm doing something off-screen. I tried to limit the use of the second screen when something was happening on the terminal for example. But sometimes I was reading the documentation off-screen and that's not good.
The chat wouldn't stay on the screen. If OBS is not the active window, the chat disappears. I need to find a way to keep it on the screen.
I should probably hype the stream a little bit more. Because streaming live with no one watching is a bit useless.
I forgot to use Presentation Mode on Phpstorm.
I don't like seeing the Dock and the Menu bar on the stream. I should probably go fullscreen.

The ugly

My MacBook Pro was really slow! I already knew that it will blow up when I start streaming but didn't expect it to be that bad.

It was so slow that every character I typed was displayed with a delay. Awful

PhpStorm was constantly indexing the project because it could not keep up with the changes. Composer and NPM were so slow that I was bored by my own stream.

My CPU had such a high load that some frames were dropped.

The sound of the fan was so loud, it was really annoying. Don't think we can hear it on the stream thanks to the AirPods Pro.

Bottom line, I have a performance problem.

What's next ?

I still want to livestream and since the app I'm building is very low priority I will continue to work on it during streams. This is a nice little project to share because I can easily explain what I'm doing along the way and can be valuable to some people.

The question is how ? I was already aiming to buy a new Mac this fall. Maybe that's the solution. But I wonder, how do other streamers do it ?

Is it a hardware problem ? A software problem ? A configuration problem ? I don't know. I will have to investigate.

Anyway, if you know anything that could help me, please let me know. I'm all ears.

How I organise my notes in 2024

2024-07-17T00:00:00+00:00

My notes always seemed messy, even to me. And then I realized they are actually organized in a certain way that pleases my brain.

Everyone is searching for the perfect notes app, the ONE notes app that will rule them all. But for me, and for as long as I can remember, I always had several different apps. In my Android days, I had 3 notes apps on my phone that had the same exact features.

Why, you ask? Because I like to put similar thoughts into their own boxes. This way my brain knows in which box to look when I need to find a thought on a certain subject.

Every thought has its place in a different app, and here is my system.

Project Ideas
Open Source Package
Tasks on a Big Project
Blog Posts
Everything Else
Conclusion

Project Ideas

I have a lot of ideas. More than I can or will code in my life. And for most of my life, I just put those ideas everywhere. In a .txt file on my desktop, on my notepad, on my phone. Literally everywhere.

Last year I decided to organize them because it became overwhelming to have so many ideas. I also discovered some nice old ideas I had totally forgotten about on an old disk.

This article from Andrew Schmelyun helped me find a system to organize my code projects and ideas.

I have four columns for my ideas: Backlog, Promising, Working On, Monitoring, and Archived. If I think of an idea, I create a new card in the Backlog and dump as much info as I can into it. Framework to use, pain points that I could encounter, potential domain names, you get the picture.

Then, I do nothing.

Basically, I use Trello to organize my ongoing and future projects into columns. Unlike Andrew, I put the minimum information there: just a name or a sentence.

If the idea is growing in my head with some details it should include or a tech stack it should use: I put these additional notes in Apple Notes in a directory named Projects where there is only one note per project.

And that's it.

Open Source Package

I don't have that many open source projects, or should I say not big enough ones. In a livestream, I saw Brendt using Github Issues as notes for his projects, checkout tempest framework.

And I think it's the best place to put those bugs & ideas: next to the package.

Tasks on a Big Project

If you don't know, my biggest project in progress is unolia.com where I'm making the best place to manage your domains.

My notes are separated into two places:

If I'm working on a feature and I need to dump new ideas on paper: Apple Notes. Because it's the fastest way to get back some RAM in my head!
If I have more than 5 seconds: Linear. Everything is eventually put there. I use priority sometimes to find important tasks at a glance. But most importantly, I use cycles to remember to take breaks and to take a step back.

Sometimes I also use TODO comments in my code. Either because I need to remember to do this before committing and if I forget Phpstorm will tell me just before. Or either because I intend to rewrite or refactor something later and it makes sense to put this note next to the concerned code.

Blog Posts

An easy one, every time I have a blog post idea, I just create a draft.

Sometimes I never write the article, sometimes I write it a year later, and sometimes, being on the admin panel of my blog pushes me to write it now.

I use to note how-to tutorials when I learn something new. It's cool because when I try to recreate something I did before, I know I have a tutorial ready here. Most of the time, I put the article online when I need to reuse my article. By having to read it again, it gives me the motivation to finish it with examples and screenshots so it can be useful to others.

Everything Else

For the rest: Notes on my phone (and my Mac). It's synced between my devices. I have some directories to separate different subjects: project, personal, unolia.com, etc.

Conclusion

The short version of this article is: put your notes where they belong.

I'm not saying you should use my system. I'm sure it doesn't work for a lot of you but I wanted to share mine because I had one for years and I just didn't notice. Hope you found this article interesting!

Help your users verify their email with Sniper Link

2024-04-19T00:00:00+00:00

Hey! I discovered this tool while using Mailcoach and I think it's very useful. May I introduce Sniper link, a little javascript tool showing a button that directly opens the user’s email client.

Here is an example of implementing it in Laravel Fortify, find verify-email.blade.php, and add the following code

Copied!

<x-guest-layout>
    <x-authentication-card>
        <x-slot name="logo">
            <x-authentication-card-logo/>
        </x-slot>

        <div class="mb-4 text-sm text-gray-600">
            {{ __('Before continuing, could you verify your email address by clicking on the link we just emailed to you? If you didn\'t receive the email, we will gladly send you another.') }}
        </div>

        @if (session('status') == 'verification-link-sent')
            <div class="mb-4 font-medium text-sm text-green-600">
                {{ __('A new verification link has been sent to the email address you provided in your profile settings.') }}
            </div>
        @endif
        <div class="flex items-center justify-around min-h-14"> 
            <sniper-link
                recipient="{{ auth()->user()->email }}"
                sender="{{ config('mail.from.address') }}"
                template="Check my emails on {provider}"
            />
            <script src="https://sniperl.ink/v1/sniper-link.js" defer></script>
        </div>

        <div class="mt-4 flex items-center justify-between">
            <form method="POST" action="{{ route('verification.send') }}">
                @csrf

                <div>
                    <x-button type="submit">
                        {{ __('Resend Verification Email') }}
                    </x-button>
                </div>
            </form>

            <div>
                <a href="{{ route('profile.show') }}" class="underline text-sm text-gray-600 hover:text-gray-900 rounded-md focus:outline-none focus:ring-2 focus:ring-offset-2 focus:ring-primary-500">
                    {{ __('Edit Profile') }}</a>

                <form method="POST" action="{{ route('logout') }}" class="inline">
                    @csrf

                    <button type="submit" class="underline text-sm text-gray-600 hover:text-gray-900 rounded-md focus:outline-none focus:ring-2 focus:ring-offset-2 focus:ring-primary-500 ms-2">
                        {{ __('Log Out') }}
                    </button>
                </form>
            </div>
        </div>
    </x-authentication-card>
</x-guest-layout>

I added a fixed height because I don't like the content to shift after initial render.

It's easy, it's automatic!

If you wanna learn more about the confirmation rate with Sniper Link, this blogpost goes deep into the subject.

Improved handling of 404 errors with Livewire

2024-02-11T00:00:00+00:00

I like to use the findOrFail() function in Livewire methods that call actions. But I really don't like the default behaviour of Livewire showing a modal view with a 404. Let me show you how to personalise that.

This is a short solution, simply copy this code into your app.js (or in a <script> tag if you want different behaviour through your app) and customize it with a modal or notification for your user. What makes the most sense for your application.

Since I'm using it in a CRUD context here, here is a simple and effective code.

Copied!

document.addEventListener('livewire:init', () => {
    Livewire.hook('request', ({ fail }) => {
        fail(({ status, preventDefault }) => {
            if (status === 404) {
                // Your preferred behaviour
                if (confirm('Ressource not found. Reload the page maybe?')) {
                    window.location.reload()
                }

                // Prevent Livewire default behaviour 
                preventDefault()
            }
        })
    })
})

And that's basically it! If you wanna learn more on the subject, checkout the Livewire documentation on Javascript hooks

As a little bonus/advanced way, here's how you can recharge just the component instead. I don't guarantee this will work in all applications, as it is a hacky way to access the Livewire request payload. Livewire can send multiple commands at once and this code may not access the correct one. This works for a full-page component with a method call, but it's a risky method and can break with a Livewire update.

Copied!

if (confirm('Ressource not found. Reload the component maybe?')) {
  let object = JSON.parse(payload);
  let snapshot = JSON.parse(object.components[0].snapshot);
  let component = Livewire.find(snapshot.memo.id);
  component.$refresh();
}

Have a nice day!

How to have transition working with dynamic list on Livewire

2024-02-11T00:00:00+00:00

If you have tried to use wire:transition in a loop in Livewire, you may have seen some weird behaiour. It's because of limitation recently documented on Livewire. But no worries, I have a workaround for you!

The solution is simple, we are gonna use AutoAnimate.js and to be more precise alpine-auto-animate wich is a Alpine.js wrapper.

Installation
Using auto animate with Livewire

Installation

The easiest way is to use the cdn to install the library

Copied!

<script src="https://cdn.jsdelivr.net/npm/@marcreichel/alpine-auto-animate@latest/dist/alpine-auto-animate.min.js" defer></script>

But I didn't go the easy way, I prefer to package all in my own project. So we will have to follow Livewire documentation on manually bundling Livewire and Alpine, then add alpine auto animate plugin.

Your app.js will now import Livewire and Alpine manually so you can add one or more plugins before Livewire start.

Copied!

import { Livewire, Alpine } from '../../vendor/livewire/livewire/dist/livewire.esm';
import AutoAnimate from '@marcreichel/alpine-auto-animate';

Alpine.plugin(AutoAnimate);

Livewire.start()

And your blade layout should include two new blade tags

Copied!

<html>
<head>
    @livewireStyles 
    @vite(['resources/js/app.js'])
</head>
<body>
    {{ $slot }}
 
    @livewireScriptConfig 
</body>
</html>

Using auto animate with Livewire

So before, you probably tried something like that

Copied!

<ul>
	@foreach($list as $model)
    	<li wire:key="{{ $model->id }}" wire:transition.duration.175ms>
        	{{ $model->name }}
        </li>
    @endforeach
</ul>

Remove wire:transition from the loop and add x-auto-animate attribute to the parent.

Copied!

<ul x-auto-animate.175ms> 
	@foreach($list as $model)
    	<li wire:key="{{ $model->id }}"> 
        	{{ $model->name }}
        </li>
    @endforeach
</ul>

Full documentation on how to cusotmize annimations on Github.

And voilà!

Eser DENIZ / @SRWieZ

Linux suspend on a T2 MacBook Pro: what actually worked

#What I tried

#Trying t2bce

#Hibernate did not save me

#What actually works

#The conclusion

Is Starlink Worth It? My Switch to Better Alternatives

#Life before Starlink

#Why I tried Starlink

#The problems piled up

#My savior: Forfait Free Max — 20€/month

#Bonus trick: Devolo Magic 2

I made 4 plugins for NativePHP

#The plugins

#nativephp-mobile-screen (free)

#nativephp-mobile-contacts (paid)

#nativephp-mobile-calendar (paid)

#nativephp-mobile-screenshots (paid)

#Why paid?

#What's next

Devkeepr: your trusted companion for your dev environment

#What's Devkeepr?

#What it does

#Why we built it

#Try it

I made a CLI tool to keep an eye on your dependencies: Whatsdiff v2 is out!

#What's Whatsdiff?

#What It Does

#How to you use it?

#The real cool stuff

#Terminal UI

#Model Context Protocol (MCP) Server

#Conclusion

Your small VPS is almost full, clean up your journal logs

#The problem

#Immediate fix

#Permanent fix – Limit current log growth

Laravel Nightwatch: Log outgoing requests from Saloon

Get the last login date of a user in Laravel

#The code

#Some notes

How to automatically control the Raspberry Pi fan with Home Assistant

#Hardware Setup

#Software Requirements

#Configuration

#Result

How I wasted 3 days upgrading to FluxUI

#Before the migration

#The big refactor

#What’s next ?

#Before and after screenshots

#Conclusion

#TLDR

Host your own AI Server using Proxmox and Ollama and connect PHPStorm to it

#Create a new VM

#Install required Nvidia CUDA Drivers and Toolkit

#Docker Installation

#Install Nvidia Container Toolkit

#Use Ollama for running AI models

#Basic Usage of Ollama

#Make Ollama listen outside of localhost

#OpenWebUI Installation

#Tailscale

#PHPStorm Integration

#Update Ollama and OpenWebUI

#Conclusion

2024 : Year in review

#Traveled a lot

#Started my own company

#Wrote a lot more

#I got payed for Open Source work

#Now I'm poor

#But somehow happy

#What's next then?

My take on Bluesky

#Mainstream

#Discover VS Following feed

#Sharing a link

#→ TLDR

What I tried

Trying t2bce

Hibernate did not save me

What actually works

The conclusion

Life before Starlink

Why I tried Starlink

The problems piled up

My savior: Forfait Free Max — 20€/month

Bonus trick: Devolo Magic 2

The plugins

nativephp-mobile-screen (free)

nativephp-mobile-contacts (paid)

nativephp-mobile-calendar (paid)

nativephp-mobile-screenshots (paid)

Why paid?

What's next

What's Devkeepr?

What it does

Why we built it

Try it

What's Whatsdiff?

What It Does

How to you use it?

The real cool stuff

Terminal UI

Model Context Protocol (MCP) Server

Conclusion

The problem

Immediate fix

Permanent fix – Limit current log growth

The code

Some notes

Hardware Setup

Software Requirements

Configuration

Result

Before the migration

The big refactor

What’s next ?

Before and after screenshots

Conclusion

TLDR

Create a new VM

Install required Nvidia CUDA Drivers and Toolkit

Docker Installation

Install Nvidia Container Toolkit

Use Ollama for running AI models

Basic Usage of Ollama

Make Ollama listen outside of localhost

OpenWebUI Installation

Tailscale

PHPStorm Integration

Update Ollama and OpenWebUI

Conclusion

Traveled a lot

Started my own company

Wrote a lot more

I got payed for Open Source work

Now I'm poor

But somehow happy

What's next then?

Mainstream

Discover VS Following feed

Sharing a link

→ TLDR

Flat file is awesome

Antlers is cool, but...

Pricing and market needs

Statamic should definitively have cloud hosting

TL;DR

Prerequisites

The problem

The solution

Customize Filament

Create a new helper class 🎨

Use it on your project

Security concern

How can we do more?

Subject of the stream