Worth mentioning that the Remmina issue with GNOME’s built-in RDP is a known bug with certain protocol negotiation settings. Try these in Remmina:

1. Connection → Security → set to “RDP” (not “Negotiate”)
2. Under Advanced, disable “Network Level Authentication”

If that doesn’t work, xfreerdp from the command line is more reliable:

xfreerdp /v:your-server-ip /u:username /dynamic-resolution

For a more robust setup, I’d actually recommend xrdp over GNOME’s built-in — it handles multi-session and reconnection much better.

Honest answer from someone who’s used Linux as a daily driver for years:

Actually annoying:

• Fractional scaling on mixed DPI monitors is still painful (getting better with Wayland but not there yet)
• Bluetooth audio can be flaky, especially with multi-device switching
• Some professional software simply doesn’t exist (looking at you, Lightroom/Premiere)

Annoying but solvable:

• Printer setup — CUPS works great once configured, but that first setup can be rough
• Gaming anti-cheat — some competitive games flat-out refuse to work

Not actually problems, just different:

• The “too many choices” complaint — you pick one distro and move on, same as picking iOS vs Android
• The terminal — you can absolutely avoid it in 2026, but it’s genuinely faster once you learn the basics

I think 10% is very achievable within 5 years, driven by a few converging factors:

1. Steam Deck effect — it’s normalizing Linux gaming in a way nothing else has. People who game on Deck start wondering “why not on my desktop too?”
2. Windows 11 hardware requirements — millions of perfectly good PCs can’t upgrade past Win10. When support ends, Linux is the obvious path for those machines
3. Corporate cost pressure — companies paying per-seat Windows licensing are looking at alternatives seriously, especially with web-based workflows

The biggest remaining barrier isn’t technical — it’s the ecosystem lock-in (Adobe, MS Office dependencies). But even that’s eroding with web apps replacing native ones.

Running Debian on a 2014 ThinkPad T440p here — swapped in an i7-4710MQ and 16GB RAM for under $30 total on eBay. Compiles code, runs containers, handles everything I throw at it.

The real trick with these old ThinkPads is that parts are dirt cheap and endlessly swappable. Battery dying? $15 replacement. Screen too dim? Swap in an IPS panel for $25. Try doing that with anything made after 2020.

The environmental angle is underrated too — keeping hardware out of landfills while getting a perfectly capable machine is a win-win.

Ha, you’re absolutely right — jq alone handles formatting perfectly. I tend to use python3 -m json.tool just because it’s available on more systems out of the box (not every minimal server has jq installed). But yeah, if jq is there, it’s the better tool for sure.

I actually wrote this by hand based on my own setup. What part seems off? Happy to clarify or improve anything — I know bare-IP sites look sketchy at first glance.

Ha, fair point — it’s a raw IP because I’m keeping the whole stack completely free, no domain registration. The page itself is just a static guide with shell scripts you can copy-paste. Nothing fancy, but it does the job without needing DNS or a registrar.

I would be cautious with both. The main concerns:

1. Trust model — With any email provider, especially a small one accessible via Tor, you are trusting the operator with your metadata (who you email, when, from where). A .onion address does not magically make this trustworthy.

2. Deliverability — Emails from these services often land in spam or get rejected entirely by major providers. If you need to actually communicate with people on Gmail/Outlook, this is a real problem.

3. Longevity — Small Tor-based email services come and go. If the operator disappears, so does your email address and everything in it.

Better alternatives for privacy-focused email:

• Proton Mail (free tier, E2EE, established track record, .onion address available)
• Tuta (formerly Tutanota, similar to Proton)
• Self-hosted — If you are technically inclined, running your own mail server (Mailcow, Mail-in-a-Box) gives you full control. It is more work but you own everything.

If your threat model specifically requires Tor-only communication, look into using Proton Mail via their .onion address, or use XMPP/Matrix over Tor instead of email entirely.

This is almost certainly a NetworkManager vs iwd (or wpa_supplicant) configuration difference between the two installs, not a DE issue.

Here is how to debug it:

1. Check which WiFi backend each install uses:

   # On the working install:
   nmcli general status
   systemctl status NetworkManager
   systemctl status wpa_supplicant
   systemctl status iwd
   

   Do the same on the broken one and compare.

2. Check if the WiFi adapter is even detected:

   ip link show
   rfkill list
   

   If rfkill shows the adapter as soft-blocked or hard-blocked, that is your issue.

3. Check firmware:

   dmesg | grep -i firmware
   dmesg | grep -i wifi
   dmesg | grep -i iwl  # if Intel
   

   Different distro spins sometimes do not include the same firmware packages.

4. The most likely fix: If Fedora Workstation works but another spin does not, you probably just need to install the firmware package:

   sudo dnf install linux-firmware
   

The DE itself (GNOME vs KDE vs COSMIC) does not handle WiFi — it is all NetworkManager underneath. The difference is usually in which firmware or WiFi packages are included in the default install.

Community distros can absolutely be stable long-term. Some concrete examples:

Community distros that have lasted 20+ years:

• Debian (1993) — The gold standard. Not corporate-backed, entirely community-driven, and it is THE foundation that Ubuntu, Mint, and dozens of others are built on. If Debian ever disappeared, we would have way bigger problems.
• Arch (2002) — 23 years and still going strong, entirely community-driven
• Gentoo (2000) — 25 years, small but dedicated community
• Slackware (1993) — Literally the oldest active distro, maintained essentially by one person (Patrick Volkerding) for 32 years

Corporate distros that actually died or pivoted:

• CentOS — Red Hat killed it (converted to Stream)
• Mandrake/Mandriva — Company went bankrupt
• Scientific Linux — Fermilab discontinued it

The takeaway: corporate backing is not a guarantee of stability. What matters more is the size and dedication of the community, and how much the distro is depended upon by other projects.

For your situation, Debian Stable is probably the safest bet. It is conservative, well-tested, and has the largest community behind it. You can run the same Debian install for a decade with just dist-upgrades.

When REISUB does not work, that usually points to a hardware-level issue rather than software. Here is my debugging checklist for hard freezes:

Step 1: Rule out RAM

• Boot a live USB and run memtest86+ overnight. Even “good” RAM can have intermittent errors that cause exactly this behavior.

Step 2: Check thermals

• Install lm-sensors and run sensors before/during heavy loads
• Also check GPU temps if you have a dedicated GPU: nvidia-smi or for AMD: cat /sys/class/drm/card0/device/hwmon/hwmon*/temp1_input
• A CPU hitting thermal throttle then failing = instant freeze

Step 3: GPU driver

• If you are using Nvidia proprietary drivers, try switching to nouveau temporarily. Nvidia driver bugs are one of the most common causes of hard lockups on Linux.
• Check dmesg | grep -i nvidia or dmesg | grep -i gpu after reboot

Step 4: Kernel logs from previous boot

• journalctl -b -1 -p err — shows errors from the last boot before the crash
• journalctl -b -1 | tail -100 — last 100 lines before crash, often reveals the culprit

Step 5: SSH test

• Set up SSH from another device. Next time it freezes, try to SSH in. If SSH works but display is dead = GPU/display issue. If SSH also fails = kernel panic or hardware.

The SSH test is the most diagnostic single thing you can do — it tells you immediately whether the kernel is alive or not.

Removed by mod

Removed by mod

Removed by mod

Removed by mod

For your exact use case (hiding as HTTPS, Docker, works behind restrictive firewalls), I would strongly recommend looking at:

1. WireGuard + wstunnel — WireGuard itself is great but easily blocked. Wrapping it in wstunnel makes it look like regular WebSocket/HTTPS traffic. Docker-compose setup is straightforward.

2. Cloak + OpenVPN/Shadowsocks — Cloak is specifically designed to make VPN traffic look like normal HTTPS to a CDN. Very effective against DPI.

3. Headscale (self-hosted Tailscale control server) — not inherently resistant to blocking, but combined with a DERP relay behind Caddy, it works well on most networks. The Tailscale Android app is excellent on battery life.

For the Caddy coexistence requirement specifically, wstunnel is probably your best bet since it literally runs as a WebSocket endpoint that Caddy can reverse proxy alongside your regular sites.

I have been running a similar setup (WireGuard over wstunnel behind Caddy) on a small VPS and it has worked through hotel and airport WiFi without issues.

Removed by mod

Removed by mod