klankin

Can’t mount networked filesystem’s in android, but you can use a file browser app to emulate the same thing.

You can mount /var and /tmp to the ssd, lot of tutorials on doing this for Pis SD cards if your googling.

The only substance I can see to it is when do you draw the line from a modified Debian (or Ubuntu) setup to a “new” distro?

If you start with an Ubuntu image its technically possible to ship of Theseus it right into an Arch image, but you could argue the default config of both is best representative of the actual distro maintainers goal (even if irrelevant to power users).

(Saying this all as a NixOS user with a system that hardly even looks like Linux sometimes so maybe I’m a bit biased on how blurry all the lines are lmao)

SElinux blocks this for aosp and its forks.

I largely agree, but I could see having a prebuild iso/img including uboot for most common boards being a lot more user friendly than doing it by hand.

That and a binary cache could make things take a couple mins for a download vs a couple days to compile the kernel + all packages for any user with lower end hardware.

Kinda like what armbian provides for the arm space, but with a lot harder initial curve by hand rolling their own distro.

Modular custom single-program kernel running in a VM live migrated across a cluster?

No no the fearmongering headline says “Linux root” it must be a kernel issue, not some obscure unnecessary meta package manager.

A/B testing, dont worry you’ll get it son enough.

Sounds more like everyone’s privacy rights are restricted just in case they might perhaps be capible of doing something dangerous in public.

Whoops, autocorrect strikes again

Mid-range networking equiptment common in higher end homelabs or small/medium enterprises.

Doesnt compete with fancier Cisco gear, but has an easy to use interface that can scale fairly well.

Though like most networking equiptment the hardware is dirt cheap, so Alpine’s lightweight base fits it well.

Most ubiquity equipment is alpine I believe

We dont yet have proof AI can “imagine” new things, just interpolates between existing. For complex relationships such as realistic fluid/particle dynamics it also requires billions of inputs before approximating reasonable outputs - so the cost to potentially nonexistent ROI timeline just doesnt add up. Its made even worse if youre already simulating billions of viable simulations, just to generate thousands.

This is why most modern techbro AI requires massive internet piracy, without already having the training data readily available (but not efficiently simulated) the algorithms arent worth much.

Tangentially this is why such algorithms have many applications in the medical field, they generally have access to a large dataset of human annotated diagnosis that can’t readily be created by a computer.

Mind linking the relavent simplex SMP/xftp windows container then?

Docker desktop for Windows runs under WSL or Hyper-V, both being VMs themselves.

Arguably running a Linux VM themselves will only offer them more customisation options (although may be heavier than WSL)

Thats how docker runs “natively” on windows, its kernel has no support for namespaces nor cgroups that containers require