Why stop at Windows? There is so much other Microsoft crap to ditch.

Or XMPP or IRC or Mumble …all of which have clients that will consume less resources on your machine

That is a different story & a usability pain I can share 😅 …but assuming there wasn’t a GC, the build is cached, just needs another switch—& I am willing to take that as opposed to having the whole process running as root.

It’ll call sudo at the point it needs to at the end regardless of remote or not. There have been a couple of bugs in the past trying to run the whole process under sudo & --use-remote-sudo was always recommended as a fix.

I mean we already had the universal protocol in an extensible markup language, but the next generation decided we needed to rewrite everything in a JSON schema that isn’t as easy to extend as XMPP. It’s federated/decontralized, has many chat clients, some social media + community managing platforms (Movim & Libervia), used for video conferencing (Jitsi & Zoom), negotiation matchmaking for games (most of them), displaying friend roster status updates (Nintendo, & many other systems). This would have (& can still be) the dream instead of needing to reinvent everything.

Worth reading: https://ploum.net/2023-06-23-how-to-kill-decentralised-networks.html

Being federated isn’t the same as living in the same Fediverse. You can upvote a Lemmy post from Mastodon since they both use ActivityPub but you can’t do that with a Matrix account. There are a couple of different ActivityPub-like generalized protocols out there, but none of them are near the size of ActivityPub & Lemmy is ActivityPub so for all intents & purposes for this conversation the Fediverse here (& most places) is ActivityPub. Matrix is on an entirely different federated network & they aren’t related.

You should at least consider nixos-rebuild --use-remote-sudo switch over raw-dogging sudo.

#!/usr/bin/env dash

set -e

cd $HOME/nixcfg

# flake inputs to update
for input in nixpkgs nixos-hardware home-manager hosts; do
	nix --extra-experimental-features flakes --no-warn-dirty flake update $input
done

# rebuild NixOS
nixos-rebuild --use-remote-sudo switch --keep-going --fallback --flake $HOME/nixcfg#$(hostname)

# check for firmware upgrades
fwupdmgr get-updates

# print hard drive status info
sudo smartctl -H /dev/nvme0n1
sudo zpool status -v -x

Just please get us proper color management. Creators need accuracy & HDR is still a mess.

They don’t package LTS kernels which is pretty concerning—especially if using out-of-kernel modules that don’t always get released in lock step that could leave you with a machine that won’t boot.

Matrix isn’t ActivityPub ∴ not a part of the fediverse

Jitsi is built atop XMPP so you have a general purpose chat server already on the system.

Movim v0.28 released within the last 24 hours. It has a web UI (that is optimized for both large & small viewports), E2EE via OMEMO, OTR, or PGP (but users can choose native clients if they wish). With the NLNet funding they are extending to full video conferencing + compatibility with the Dino native GTK client. Subjectively, it looks pretty sharp for a web client. You can also use it to share ‘posts’ for announcements & public feed aggregation—something a group chat should never be used for (announcements & other long-term messages get lost in the black hole search can’t find & unreleated posts all around it with messy-to-follow threads since this sort of content isn’t supposed to be chat).

It’s not quite as easy as services.movim.enable = true for NixOS but the NixOS module isn’t far off once an XMPP server has been selected with optimized defaults beyond standard setup—& the option I would personally recommend for self-hosting as declarative config is easier to work with in the long run, but there are non-Nix options. Being PHP, it’s fairly performant as well as not being built on some space-wasting, RAM-sucking ‘eventual consistency’ model that will cost you out the ass (which is Matrix, by design). The front-end, being mostly vanilla JS, is not using some heavy, bloaty framework. This will meet all your needs & not require expensive hardware host even on an old laptop at home or part of a multi-purpose server (does not need dedicated hardware).

“FOSS” here doesn’t mean hosted exclusively on proprietary Microsoft GitHub, right?

Naïve me it college hearing non-destructive editing was right around the corner. The corner was well over a decade. 😅

There are several open protocols that meet your criteria that aren’t Matrix (with most of them using double-ratchet encryption similar to if not exactly like Signal). Due to server costs (Matrix eats a lot of RAM & storage), medium-sized entities usually bow out so the Matrix network largely consist of a few 1–10 user servers & massive centralization around Matrix.org & the hosted servers they provide. Since almost all the messages get synced to the Matrix.org server if just one Matrix.org user is in your room or whatever, Matrix have the largest access to all that metadata.

Signal refuses to even try to accommodate for UnifiedPush or MQTT for those not using play services requiring an extra battery-draining socket to their servers. You are also still required to use one of the mobile duopoly OSs as a primary device to register (SIM still required). Good luck if you use a Linux phone, KaiOS, or just don’t want an ever-present tracking beacon on you. We all know the Electron-based desktop client is shit. I would flip this on its head & say it is the service’s probably if they choose to prioritize & mainly support the shitty mobile OS duopoly it’s their problem for providing a bad service & getting the criticism they deserve.

I like it. Take my vote in your next election.

This is how they want to frame it. C has footguns, therefore use Rust—instead of Rust is one of the options you could use.

Correct me if I am wrong, but my understanding is that you use Coq to prove your theroem, then need to rewrite it in something else. I think there is some OCaml integration, but OCaml—while having create performance for a high level language & fairly predictable output—isn’t well-suited for very low-level kernel code. The difference in the ATS case (with the ML syntax similarity 🤘) is you can a) write it all in a single language & b) you can interweave proof, type, & value-level code thru the language instead of separating them; which means your functions need to make the proof-level asserts inside their bodies to satisfy the compiler if written with these requirements, or the type level asserting the linear type usage with value-level requirements to if allocating memory, must deallocate memory as well as compeletly prevent double free & use after free.

For those in the back: Rust can’t do this with its affine types only preventing using a resource multiple times (at most once), where linear types say you must use once & can only use once.