I see this regularly: "I use GitHub to backup my local repos."
If `gh repo ...` commands get run you can lose everything instantly. You can force push and be left with a single blank commit on both sides. The agent has full control of everything, not just your local data.
Just set up Rclone/restic and get your stuff into a system with some immutability.
Force pushing doesn't actually remove anything from the remote repository, only changes some references for which commits the branches point to. Plus, any forks on github will be completely unaffected. It's not perfect, since Github doesn't seem to offer any history of such reference alterations (a la the reflog), but it's still a valuable offsite backup from a developer's perspective.
Okay, fair enough re force pushing (though `gh repo delete` is still an option). I suppose for a sufficiently active codebase copies of it will exist elsewhere. Just seems odd to me that people aren't backing up anything else on their computers otherwise they could trivially just include their git-based projects.
A NetBSD posted a blog stating NetBSD is having issues porting Wayland due to Linux specific items. OpenBSD stated something similar.
Both articles indicated it will be a very long time, if ever, to get Wayland fully working on their systems. I did see this presentation that describes some issues as of 2025.
As someone about to learn the basics of Terraform, with an interest in geo-distributed storage, and with some Hetzner credit sitting idle... I came across the perfect comment this morning.
It's possible to both over-squash and under-squash. You want each commit to do one thing (conceptually), and if you make a lot of in-progress commits, you do want to squash those. But squashing a bunch of related "things" into one commit is too much. The art is in recognizing what one "thing" is.
Changed "make a thing" to "describe a tool" and got a raft of text back that I couldn't be bothered to think about, because it's not my space. This jumped out though:
"## Existing things that almost do this (but don’t go far enough)
* Radicle – closest philosophically
[...]"
So it seems to me that you succinctly described Radicle - at least well enough for an LLM to recognize it.
I think the dichotomy is more "curated commercial ecosystem of mass-appeal winners" vs "the full spectrum of human experience and perspectives that requires effort on the part of the listener to find what is worth listening to".
Personally, I prefer the latter. You get out what you put in and the best podcasts (for me) have always been passionate people trying to share things with others. There are definitely some with high production value that I would (and do) miss but they were never sustainable to begin with so nothing truly lost there.
Podcasting will always be able to endure in its most basic form: two people, a mic, and an RSS feed.
I'm not saying it's not important. It's that I don't have any control over it. Do I have enough "cache" in your opinion? Presently I have:
Used: 14.6
Avail: 16.6
Cache: 4.2
Free: 6.4
If I'm asking the question "Is my system starved of RAM?" I am going to look at USED and AVAIL. CACHE is a curiosity more than anything else and FREE tells me nothing.
Also, under what circumstances would I have a ton of AVAIL and be starving the CACHE? That's an OS/kernel problem. If AVAIL drops to zero I can expect to get OOM crashing pretty quick.
atop (one of the many variants of top) reports numbers for "free", "available", and "cache" in its total systems report. As long as "available" is high, I don't worry about the RAM.
I've been running a Jellyfin server for nearly four years now. I never tried Plex or Emby but JF has been an impressive bit of software.
The availability of clients (Roku, Apple TV, Android, Xbox) is good enough that I have no problem inviting friends and family to join mine. I've learned so much about the tech bubble I live in simply from getting them onto the server.
I think the biggest obstacle to adoption beyond simple home servers is the reliance on SQLlite. If it were possible to set it up with Postgres you could run a monster server on AWS with RDS, S3, a Kubernetes. Not sure about the business case for that... but I would enjoy setting it up and pushing it to its limits.
Reminds me of when people praise the efficiency of their car: "It's great, $30 lasts me all month!". Tells you nothing about the car and hardly anything about the person using it.
I have crashes on a monthly basis but I'm also pushing it pretty hard with 4 Activities, 5 Workspaces, and 3 Screens. That's 60 desktops for it to manage (granted, only about 10-15 in total actually have anything on them) with my tiling window management scripts on top of that. Plasmashell and Kwin take up 20% of my CPU on average which is unfortunate but that's the cost for my setup I guess.
I used to daily drive KDE up until shortly after the 4 switch, when I moved to Mac. I've moved back to Linux starting in 2018 but went with i3. I installed KDE around Christmas to try it, and while I'm mostly impressed with the general polish (except that firefox doesn't react the same way as other windows to clicking on the frame), I have a hard time figuring out what activities are and how they're different from multiple workspaces.
Speaking of workspaces, is there no way to only have it show a small rectangle per space in the taskbar instead of a big wide one (I'm using multiple screens)? That's just a useless waste of space.
Activities function best as "context domains" (the classic split is "Personal" and "Work") while preserving your existing Workspaces.
I use workspaces to group apps/tasks/programs functionally (eg. "Active Projects", background stuff like music or a long running terminal task, no tiling).
Some things like Obsidian or Spotify are open on multiple Workspaces and multiple activities at the same time but only require a single instance.
> Some things like Obsidian or Spotify are open on multiple Workspaces and multiple activities at the same time but only require a single instance.
Oh, you can do that? My first impression was that the activities were somehow completely separate, somewhat like profiles in firefox. Since I actually have stuff that isn't exactly context-related, specifically spotify, I thought it would be a pain to have to switch back and forth to interact with them.
> Speaking of workspaces, is there no way to only have it show a small rectangle per space in the taskbar instead of a big wide one (I'm using multiple screens)? That's just a useless waste of space.
I missed this one. Yeah, I have the same problem and couldn't fix it but I use a tiling window manager so I didn't need that space for anything else. I've come to appreciate the overview actually and you can drag-and-drop windows between workspaces without having to jump around.
I think the best way to describe Activities is that they filter "what" is available whereas Workspaces filter/select "where" it goes. Spotify automatically chooses my "Background Tasks" Workspace when it opens and it is available on all my Activities. My task manager always shows up in my main workspace but not in my "Gaming" activity. It's a really powerful feature once you understand how it works (eg. notifications from other activities can be muted; separate file and folder views; email accounts hidden in Thunderbird, etc.)
Edit: You could probably get total isolation by using the other TTYs! A fully separate user and different desktop environment even. You just need to use Alt+F{1..9} to cycle between them. For a specialized workstation this could actually be a clean way to handle it.
I have one for general stuff, one for when I'm working, mainly to minimise distractions, and one to do 3D work.
Each activity has 9 virtual desktops because that's what I use in the main profile.
And yeah Firefox is a bit different, however you can select an option to get the regular KDE menu bar back! I've done this.
And I configured my workspaces in a 3x3 grid, that way it doesn't take so much space (and also it's much easier to navigate them than 9 in a row!). This grid function is really one of the big things that annoys me in macOS (it used to have grids but they killed it when mission control came out), windows and Gnome. I need a grid, especially because I switch with hardware keys.
If `gh repo ...` commands get run you can lose everything instantly. You can force push and be left with a single blank commit on both sides. The agent has full control of everything, not just your local data.
Just set up Rclone/restic and get your stuff into a system with some immutability.
reply