Are you Y2Q compliant? Will is happen by April 14, 2030?

This year, I'm automating self-hosting. My goal is to have all the services we use done by the end of the year. Nextcloud, jitsi, gitlab, matrix, element, bind, djbdns, the PKI/CA server... all of it. Sone of the things I've done have already starting to get reused. It's getting easier One I've got that rock solid foundation, it'll be time to rip. 🔥

GD mates. I'll try to fill that time between you GM and GN posts with this lovely picture I snapped today. Or is it AI generated? Can you even tell?

Some people still can't tell the difference between CDN #vulnerabilities and HTTP/1.1 vulnerabilities. 🤣 I guess the ol' saying is still worth its weight in gold... "It is difficult to get anybody to understand something, when their salary depends on them not understanding it." — Upton Sinclair

HTTP/1.1 Must Die

Upstream HTTP/1.1 is inherently insecure, and routinely exposes millions of websites to hostile takeover. Join the mission to kill HTTP/1.1 now

I've finally started to get some intuition with #3dPrinting issues. I was at our local #makerspace today and the Lulzbot Mini 1 was failing to bed level. I watched it closely snd saw the nozzle never made context with the corner pad. I though, "it doesn't seem to know where the bed is" which immediately led me to "z-axis offset wrong?" A few M851 commands later and lots of testing and I was able to get it high enough to pass the leveling process, but low enough to maybe print okay. We'll see. I also think I understand how to fix it so the nozzle passes the leveling process but isn't so high off the build plate when it goes to actually print (lowering it more causes the leveling to fail). So next time I'm in, I plan on trying to print some shims to raise the entire bed assembly. The only concern I have is the belt drive, as the shims will not raise the height of the stepper motor. It felt pretty good to be able to deduce the problem from first principles instead of guess & checking!

Scotch #Peatstr

This blog post has not aged well, or it has, depending on your perspective.

Buying eggs with bitcoins | FRED Blog

😍

Roces Sport Srl

5th Savosin Fire

I learned a fair amount about reprepro today. It's been 15 years since someone requested the feature to have multiple versions of a package in their repo. Patches to implement that feature arrived more than 10 years ago. I believe it was in an official release 6 years ago. It's still not in Debian stable, testing, nor sid. It is available in the EXPERIMENTAL repo and it worked great. Just can't seem to get the new version into testing, sadly. It mirrors (ha, pun intended!) my experience with trying to get a patchlevel bug fix into pamu2f. I spent a year submitting pull requests, following up on the tickets and pull request, tracking down people and emailinb them, finding people on IRC, trying to get a mentor so I could become a volunteer contributor... and in the end all that work died on the vine. Eventually someone got a minor version update and put it into testing directly. That's the power of having an @debian.org email address. If you don't have that, you can't really contribute to the project.

LOL at people suggesting AI is going to leave people behind if they can't afford to pay for the tools. If that's true, then open source (including the training data) is vital to innovation. There'd be only megacorporatons making software, no open source because no AI gatekeeper is going to allow someone to create a free and open source alternative to what they make. It'd undermine their income stream. And if you think AI can't make AI tools, then you are acknowledging a huge limitation of so-called AI. They can't make anything new. It can only remix and combine things it's seen before. So if you create anything novel, you needn't worry about being left behind.