For some context: @hswaw is a volunteer-ran non-profit community of tinkerers. This Supermicro is our best gear, and it supports not just our services, but those of 3 other hackerspaces. And soon, it will support @mykoryza -- a non-profit that aims to host web services (such as Fediverse) for the public.

If that's something you'd like to support, you can :)

This is a long shot, but: does anyone have access to a Supermicro H11DST-B *version 2.x* motherboard?

We (@hswaw) suspect that the only important difference with 1.x is a larger NOR flash for the firmware. If so, a little hot air action would allow an upgrade to Zen 2 CPUs, and get big perf/power-efficiency gains on a budget.

Boosts much appreciated!

An upgrade to @hswaw's k8s cluster is coming up! From 2x10Gb (SFP+) to 2x25Gb (SFP28) + 2x10GbE.

@implr found a good deal on these. Since these are a proprietary Supermicro form factor (and not PCIe or OCP), they tend to be pricey.

ok what

bijbel

I'd argue against calling software engineering "solved" (I'd use the word "disreputable" instead), but otherwise I agree with this article here.

It was already extremely disheartening to see basically the entire global software industry abandon all quality overnight. Imagine you run a restaurant out in the countryside and you have your regular customers, until one day some company puts up a dispenser on the other side of the road that pumps out some undefined, brown, lukewarm broth that tastes like cardboard but only costs 20% of a meal at your restaurant, and you have to watch as all but maybe one client never set foot in your restaurant ever again.

But it's much worse when it comes to things like blog posts. I have a list of technical topics that I once would've liked to do research on and write blog posts about... but what's the point now? I would only be feeding a bullshit generator so that it could generate some slightly more plausible bullshit in the future, making it yet more effort to debunk. And if any technical opportunity arose from it, then at least 3 slop projects would pop up immediately, in turn demotivating anyone who'd work on any reliable implementation (and their motivation had already taken a serious hit because they no longer consider the operating systems reliable that their software was supposed to run on, and it's a considerable amount of work just to remain where they are...).

And yeah, this year I've also noticed a distinct lack of... idk, "interesting technical things"? There was that one blog post from the FEX devs, that was really interesting. Then there was some good news about battery technology research (both lithium and sodium), which I assume will have a serious impact in the real world, but on an intellectual level, as someone who's not in that field at all, I'd file those under "neat". And that's it. That's the entirety of "interesting creative technological output" that I remember coming across in 2026.

https://borretti.me/article/no-man-is-an-island

[As Pablo Picasso said of computers]: “But they are useless. They can only give you answers.”

Felt this strongly after talking of an quaint concept (parallels between atomic orbital structure and spherical harmonics in CG rendering) to a buddy a few days ago... only to be asked why I haven't queried an LLM for an answer. Damn it, I don't want answers, I want curiosity and whimsy and science and engineering and art and beauty!

every conversation about the potential usefulness of AI, divorced from ethical concerns, is just this dril tweet

Today: fractal rendering on sgi Origin300 rack using arraysvc in sgi-tess by supernova.

what if BhutanesePassport.oga was recorded on a karaoke machine at a bar, that's why it has autotune and reverb?

me patiently waiting for yet another 'sorry, out of scope' VRP submission result

the design, if anybody had forgotten, screenprinted on black

God this whole thing makes me want to get myself lobotomized.

I guess it's time to learn FoundationDB?

Apple is too busy making billions of dollars to update its website... Wonder how their Git repo is doing!

*looks at GH username ending in -oai and a squished avatar because of course GH is now just slop*

Oh no.

Hey I wonder what's the state of distributed databases in 2026... Let's check out their websites!

CockroachDB: “Resilient AI scale without compromise.”
YugabyteDB: “Distributed PostgreSQL for the Agentic Enterprise”
ScyllaDB: “ScyllaDB for Real-Time AI”
Clickhouse: “The leading database for AI”
TiDB: “AI-Native Distributed SQL Database”

Extremely normal industry.

This really seems to have gotten significantly worse in the past couple of months

It's bad enough that I am genuinely tempted to just ask a clanker instead, but I'd rather ween myself off search in general and instead go back to reading relevant sources directly. I usually know where the thing I'm looking for is, I just got lazy when web search was good...

I know we've all been complaining that Google web search has been getting worse for years...

But it's completely useless now, right? I feel like any attempt to look for any reference data returns completely unrelated results? And it's not just LLM spam, some of it is 'old' content, just not the right result for the query?

Eg. "go net http reverse proxy" returns a link to Go's stdlib source, 4 blogspam-ish results and only then a link to godoc. I'm sure that used to be first a while back...

IT'S BACK

the news of its demise were grealty exaggerated. truly the isp that refuses to die. we love you openfactory!

world's loneliest router