_kb

How do you know it hasn't already happened?

> why does that need to be an OS?

It doesn't. You can still run Incus on other platforms of choice.

Sample size 1 here, but a big advantage of the 'full-stack' approach is things like network config, storage management, boot safety etc all work out of the box and you then get a single API (and nice client) for the whole machine. I get the benefits of cloud infra, like not having to care (too much) about sysadmin, from some hardware sitting in the corner.

I can literally

    incus launch images:nixos/unstable foo -t aws:m1.large

and start hacking.

Previously I would still need to be maintaining that base layer too. That still makes sense for some environments, but particularly for home I just want my lights and music to work, and be able to play.

Those auth UX challenges are being worked on:

https://github.com/lxc/incus-os/issues/496

https://github.com/lxc/incus-os/issues/497

IMO the client certs are pretty elegant from a technical perspective. It works well with the CLI, but the browser experience is different enough to cause at least some base level wtf-ery.

There's a bit more of a conversation / demo here which is pretty impressive: https://www.youtube.com/watch?v=KbZfbqHeJNU.

+1 -- consistency is key.

Wezterm runs everywhere, but lets me customise it once and keep that config uniform across all machines.

I can have a single config [0], wrap that in a nix expression [1] for anywhere that runs home-manager / NixOS and then also check it out and symlink on Windows machines as my portal to WSL. As my preferences change, my tooling stays consistent and familiar everywhere it's needed.

[0]: https://git.sr.ht/~kb/env/tree/main/item/dotfiles/wezterm.lu...

[1]: https://git.sr.ht/~kb/env/tree/main/item/programs/wezterm.ni...

Or more abstractly: post anything to the internet and people will always detail how you’re wrong. Sometimes that can be useful.

And of course the `olfactoryFactory` component for Enterprise™ frameworks.

How very kafkaesque.

This runs (for free) across all payphones in Australia each year: https://www.telstra.com.au/exchange/how-we-re-helping-santa-...

My tiny human loves it. I think they’re almost old enough to start learning the joys of jailbreaking this year as a modern twist on phreaking.

I don’t at the moment. This setup is new and my current hot storage needs are pretty minimal so I’m all in on NVMe. When that changes though thats the expansion plan. ASM1166 based boards seem to be an ok choice, but don’t have any personal recs there (yet).

There's a bit of a trend of vendors packaging mobile CPUs in desktop form factor which are a good candidate for this. Rather than the prebuilt mini PCs this also includes mini-ITX boards. Personally I use the Minisforum BD795i SE, but there are others too.

Check for PCIe bifurcation support. If that's there you can pop in a PCIe to quad M.2 adapter. That will split a PCIe x16 slot into 4 x M.2s. Each of those (and the M.2s already on the motherboard) can then be loaded with either an NVMe drive or an M.2 to SATA adapter, with each adapter providing 6 x SATA ports. That setup gives a lot of flexibility to build out a fairly extensive storage array with both NVMe and spinning platters and no USB in sight.

As a nice side effect of the honestly bonkers amount of compute in those boards there's also plenty of capacity to run other VM workloads on the same metal which lets a lot of the storage access happen locally rather than over the network. For me, that means the on-board 2.5GbE NIC is more than fine, but if not you can also load a M.2 to 10GbE adapter(s) as needed.

Outside of AV1/2 (and linear media in general) that's already well and truly developed tech. Nvidia DLSS, AMD FSR and Intel XeSS all provide spatial/temporal super sampling to process lower fidelity base renders [0].

There also seems to be a fair bit of attention on that problem space from the real-time comms vendors with Cisco [1], Microsoft [2] and Google [3] already leaning on model based audio codecs. With the advantages that provides both around packet loss mitigation and shifting costs to end user (aka free) compute and away from central infra I can't see that not extending to the video channel too.

[0]: https://mtisoftware.com/understanding-ai-upscaling-how-dlss-...

[1]: https://www.webex.com/gp/webex-ai-codec.html

[2]: https://techcommunity.microsoft.com/blog/microsoftteamsblog/...

[3]: https://research.google/blog/lyra-a-new-very-low-bitrate-cod...

*Richard Dawkins

The God Delusion, by Richard Hawkins.

That assertion is very location dependant.

It's certainly common across a lot of Australia, including libraries running dedicated facilities for co-working and makerspaces. As an example: https://www.slq.qld.gov.au/visit/spaces/edge/bookable-spaces.

Both state (https://precinctqld.com.au/) and local (https://businesshub.brisbane.qld.au/coworking) governments then also run free or heavily subsided workspaces if you need something less ephemeral too.

If xkcd took comment form, this would be it.

“The laws of mathematics are very commendable, but the only law that applies in Australia is the law of Australia.”

https://www.newscientist.com/article/2140747-laws-of-mathema...

All Australians now live with the Assistance and Access Act 2018, where yes in fact if you use the illegal math, receive a TCN and do not comply… straight to jail.

Just have the detector always return true. You’ll likely be within acceptable error bounds.

For those interested in an extension from this article, https://howthefouriertransformworks.com/ has a beautifully kitsch set of videos on the topic.

They do a really good job at breaking down the fundamental knowledge needed to build an understanding.

I don’t think having them stand under low flying aircraft is much safer.

You just need to scale out more. As you approach infinite monkeys, sorry - models, you'll surely get the result you need.

> So what comes next?

Feudalism.

Or for something likely relevant to many here - 3 point time estimates for project planning.

Probability distributions (even very simple ones) provide a much clearer view across any domain where there’s inherent uncertainty.

Stand next to it with a chocolate bar in your pocket. I believe that’s the canonical way to check if it’s active.

With a few hacks, you can already get web content rendered with HDR: https://fff.kidi.ng/.

It's the big flat one at the bottom.

You can pull the nix logs from here: https://github.com/NixOS/infra/blob/main/metrics/fastly/READ...

Could be an interesting data source to explore that opinion.

I don’t think that’s correct. The dot operator is composition with the same semantics as when using normal math notation.

   (f . g) x

is equivalent to

   f (g x)

The lock in is issues, pull requests, and all the conversations that shape long term software projects. The actual implementation and what's tracked is often just what falls out of that.

Ah yes, a sudomasochist.