fly + tailscale/tsnet in the container gets close, but I haven't found a way to do that and retain the "scale to zero"/"wakeup fast on request" behaviour that makes fly so useful in the first place when tailscale is in use.

native integration to make that work would be amazing.

insulation - noise and thermal.

Great post! IMO the quantization stuff is good (concrete demo of why the loss of precision is ~OK) .

For me, the refresher on FP representation was the highlight. Even having learnt it supposedly getting my CS degree, I still got caught out and learnt something.

Highly Recommended :)

what's the status or ETA on SOC2/IAM controls to support business/SaaS use-cases?

rose.systems/animalist

108 animals listed
🐏🐄🐴🐕🐽🦜🐦🪰🪳🐘🦏🐒🐅🦁🐦🐍🦂🦛🦘🐨🦙🫎🦣🐛🪰🕷🦟🥁🦈🐋🐙🦑🦪🦞🦐🐻‍❄️𓆝🦀🐱🦜🦜🐟🦎𓆈🐛🐛🐛🐥🐦𓅨🐦🕊🐦🐦🐦🦩🐦🦅𓅃🐦🦌🦬🐰🦊🪿🦢🐐𓃥🐼🐻🦓𓃴🐧🐛🦋🐌🐛🐛🐛🐛

* I haven't worked out how to logout/login without fully restarting (e.g. to quickly test startup behaviour): systemctl restart sddm should work, but something about how hyprsets is starting Joplin + Slack makes them hang infinitely.

all just minor things, overall very happy.

* a bunch of issues with voxtype wtype based input not working reliably with cursor - switched to paste mode for now... :(
* sporadic ghosttty crashes - takes out **every** console :(
* yesterday everything *except* ghostty got terminated and I still don't know why...

omarchy specific:
* a bunch of waybar config tweaks for personal preferences
* sooo many keybindings to learn to drive hyprland usefully
* configuring desktops was a pain - I've settled on a setup with hyprsets to re-open/start all my stuff, but feels janky/took a lot of effort to get working.

arch seems like a bit of a wild west in terms of package management (yeeting random crap in via AUR is terrifyingly easy) vs the trust I have in Debian is probably the biggest + just general unfamiliarity with how to drive arch overall after 20+ yrs of Debian systems makes for plenty of rough edges!

I also switched (from Gnome) to Omarchy in Dec. It's a bit rough around the edges (reminiscent of my early Slackware intro to Linux in the 90s...) but the simplicity and ability to work out wtf is going on and fix it (compared to Gnome's increasingly obscure junk) is refreshing.

Stadium scene of people waiting for a show to start. The screen behind the stage shows point floating bubbles with a circle containing a white play symbol in the centre.

Nearly show time!

great video, banger tune.

mattb.nz's Bluesky Wrapped 2025 Check out mattb.nz's year on Bluesky!

Neat tool =>

This year on Bluesky I wrote 15 posts and 42 replies. I received 71 likes, whereas 5 was from my most popular post, and apparently I love saying "just" and ✨!

www.madebyolof.com/bluesky-wrap...

makes sense, thanks for helping to clarify my thinking on this!

If I'm understanding correctly, you're saying "pick the right tool for the job" - baseline analysis starts with premise that thing (pandemic/earthquake) was an isolated, one-off event that *should* be ignored. If you believe otherwise (my question), you should pick a different tool for the analysis?

What's currently thought to be "abnormal" might just be the start of a more variable period, etc?

not a stats expert, but genuinely interested:

Future pandemics may have similar impacts, so wouldn't the right approach be simply including the "abnormal" years?

The result being more variable is a true reflection of the world! If you massage the inputs, how do you defend the conclusions?

oh my god it's incredible

between this and nanochat, I'm very tempted to just bunk off work for the rest of the year and dive in....

stupid responsibilities!

I look forward to the post I assume this is leading to for further enlightenment - I've been long puzzled by how seemingly simple/small the code around the giant pile of weights seems to be in practice, these sorts of insights are awesome.

huh, this (cache value, being a pure functional mapping from the input prompt) is a TIL moment for me...

I naively assumed that the point of prompt caching was about restoring internal state of the model...!

In hindsight that assumption seems obviously dumb, given the sizes involved!

Thanks :)

But IMO just as easy to put the server in a secured room with badge access logs, etc, and/or a locked rack in the corner of the office vs under a desk.

I don't think a compromised dev machine is comparable - they should *never* have secrets directly exposed them (vs CI/CD which requires them)...

Yes, SOC2 in my experience is mostly validating that you have a set of policies and controls in place, that you assert are suitable for your business (vs a very low-bar baseline) and that you actually follow them.

So if you want to declare this not a risk, your auditor will probably accept it.

I'd be more worried about the security/supply chain risks:

Assumption: You sell a product to/maintain OSS used by someone important that attacker X wants to compromise.

Threat model: X breaks into your office, compromises your under-desk CI server with subtle malware that backdoors your builds.

Under-desk (vs on-prem server room) also raises physical security questions (e.g. evil maid/cleaner attack) that I would find harder to justify SOC2/ISO controls against.

A CI server is riskier than a dev desktop - it deploys directly to prod, while desktop actions are gated through a review step.

I'd look at it less from a reliability perspective and more from maintenance and security.

Under-desk might be fine if it's well-managed (updated, monitored, etc) but "spare box" has connotations that point away from that...

Is the under-desk runner in your MDM/inventory and regularly updated?

I may be featured in this meme...

Looks exciting!

watching with interesting, and intruiged by the idea, but timezones are challenging...

If/when you have an iteration of this that works for UTC+12/UTC+13 (NZ) I would be interested.

added to my queue, but do you know why the transistor share page doesn't link to Spotify?

I had to spend an extra minute manually searching for it in Spotify...