I do this with my kernel & a couple of applications that either compile super fast anyhow or at runtime benefit from further compiler optimizations.

NixOS & OpenWRT are my two. NixOS’s Nix language as declarative config is such a great tool for setting up & maintaining a machines for the long-term that despite the initial learning curve has paid off in the long run (Guix or a Nix successor should also be in the same category). OpenWRT is the purpose-built tool it is for having an OS for a router with low overhead & a UI that can be easier to understand the config when networking isn’t something you do on the regular.

Depends how you view it & how green the grass is on the other side. Personally the Forgejo approach of copying MS GitHub to ease onboarding doesn’t resonate with me as a user over, say, making a better product by fixing some of the major flaws like the pull request model being a major slowdown, CI in YAML soup, needless social features… but others prefer this approach & a rocked boat is scary.

I had one with fresh minced mint + basil along with diced bittersweet chocolate & it went so beyond the average enjoyment of mint chocolate.

Not just community managed but operated as a non-profit. Codeberg won’t be scraping your deleted history to train their LLMs that they will sell back to you unlike Microsoft.

I am still convinced Git is overrated & overly complicated—and it is a shame all of the decent forges (even basic ones) are all built around Git.

Gemini has accessibility & bandwidth problem. HTML is a more accessible format & HTTP offers compression. Add that Gemtext has too few ‘elements’ for technical writing or even basic blogging & I don’t think it should be seriously considered for anything than a novel toy.

GitHub isn’t open source

This needs to be repeated for those in the back that still didn’t get the memo. You do not need to use Microsoft products, especially if your goal is free, open, and/or ethical software.

To upgrade the UEFI or other hardware-level firmware you need a way to upgrade. Best OEMs use LFVS; good OEMs use have ISOs or bin files you can flash from UEFI; terrible OEMs lock that into a Windows-only executable.

In my case there was a fan & thermal update I was never able to get.

I find myself not often using my open hardware USB security token since it is inconvenient to use a dongle (same reason phones should never have dropped headphone jacks)

None.

I had a Razer laptop in the past when they were talking about being dev laptop forward & supporting Linux.

This never happened. Instead flashing Linux voids the warranty now, support drops you, & firmware upgrades only happen thru a green-accented genuine Microsoft Windows GUI installation (no *.bin flashing, no CLI FreeDOS support, no Windows PE).

The USB 2 is for a mouse or keyboard as extra bandwidth does nothing.

I can’t relate tho. My laptop only has two USB4 with type-C connectors. I really wish it tad at least one type-A connection. Luckily unlike my last laptop, at least this one has a headphone jack (I guess ASUS learned from their folly & never skipped the jack on future models).

services.miniflux.enable = true on NixOS is even easier

China, IL would be on my list

How on Earth do you lose this support in the handover shuffle—especially knowing the audience a third-party email client? I would say it shouldn’t be released if F-Droid support isn’t there since it isn’t something you would want to back-burner.

One important thing K-9 does that this doesn’t: realese on F-Droid.

What is this? At least provide a repository like DivestOS… while you are at it, get the code for the free software off of proprietary Microsoft GitHub.

The biggest shake-ups in a while outside we-don’t-use-X (no systemd, etc.) are the declarative distros like NixOS & Guix. You do the whole system setup & config thru a single file (or broken into multiple). Learning curve is very high for the config but the payoff is less things changing out from under you & setting up new machines & rolling back to working states without resorting to FS snapshots. They are good languages to learn for software development too where you want repeatable software.

I use the note to self capability of my XMPP server to send a message to myself for these sort of one-offs. I would never want my data in the hands of some proprietary service if I have the option—sharing data just to yourself on these services also means it is Big Tech’s data now too. All of the XMPP clients are super lightweight.

Bigger cases, I will use scp, rsync, or magic wormhole. Or just using removeable storage.

The ability to defederate arguable makes it more free & open even if it isn’t what I would prescribe.

I recall having some fun with League of Legends when you could just join chat & chat rooms thru a regular XMPP client. This was convenient at work on Linux to not need a working client to catch important messages from teammates. But everyone wants a walled garden now.

Which means your two close friends & the corporate Discord data harvester.

With friends I can at least tell them to use a privacy-respecting or self hosted option, but I still don’t understand how businesses actively prefer US-based, proprietary services such as Google Meet for meetings. Do they really think Google & others aren’t record & training on their private business meeting data?

Not open? Wat