@Waterfox hi there! Are there any plans to fix "Recent activity" on starting page? It stopped working with the release of 6.7, and it's a really nice addition, that should be there. Or if it's removed intentionally, the toggle to enable it should be deleted from the settings.
A bit tired of clunkyness of Ollama+AnythingLLM, I decided to try something new.
#LocalAI is a great piece of software. All-in-one solution for downloading models, backends, WebUI for interacting with the AI, and all of that running in Docker with GPU acceleration.
I even used "Configurable Button" Plasmoid to have an on/off switch on my panel, so I'm one click away from starting a local AI model.
And the performance! 95 tok/s on RX7900XTX
I hate AI for one particular reason:
You browse the web, you stumble upon a nice-looking FOSS project, something that would be actually very useful. You open the Github page, read the README, looks nice.
Then you look at the commits and contributors. Top contributor - Cursor Agent or Claude. You look into the code - entirely vibe coded with massive security and performance oversights.
Sigh... Yes, develop FOSS applicatios. No, don't depend entirely on AI...
Let's hope I won't jinx it, but #solus is looking and working pretty nicely on my PC. Blazing fast boot times, better framerate in VR, and eopkg is just flying, the fastest package manager I have ever seen.
@getsolus you guys are doing some wild sorcery...
#linux #foss #solus #distro #oc #gaming #plasma #kde #distrohopping
After another month of waiting for response (mBank has 30 days to respond to my complaints, so naturally they respond after 29 days) I finally understand why they block mBank PL app on GrapheneOS.
They genuinely think, that "rooted Android" means "any alternative Android ROM". So even despite GrapheneOS not granting user root permissions (so not "rooting" the device), they still consider it "rooted".
It all makes sense now! (it doesn't)
Where should I file a suggestion for change in KDE Discover?
Based on talks with new Linux users, many of them are confused when they open "Software Center" and are met with multiple "sources" for a single application - from repos, Flatpak, Snap.
I think Discover should include a small ? icon next to the source selector, explaining what specific format is, what are the pros and cons (eg. "This is repo package, it may be outdated if you're using an LTS release").
I just learned that Fedora plans to create a whole separate edition called Fedora AI Desktop. It's going to be an immutable edition, that comes packed full of tools and libraries essential for working with both local and remote AI.
And to be honest... It's the most mature and non-controversial way to "tickle" the AI-crazed enterprise space, without compromising their core product.
If it works out Fedora will have one more edition. If not, they shut it down, no biggie. But the core is safe!
So, going on a week-long trip. I was going to take both my SteamDeck and laptop, but at one point I stopped and wondered.
Why the hell would I take a laptop with me? I have basically a laptop with a gamepad attached to it. It's 100% enough for what I will need on vacation!
THIS is the power of Steam Deck. Not its gaming performance, not the price. The fact, that with a pretty decent handheld console, you get a Linux laptop free!
A small release - I decided to build an open-source digital photo frame from Raspberry Pi Zero 2 W and my local Immich server.
https://codeberg.org/leniwcowaty/PiFrame
It even has a 3D printed case!
#raspberryPi #zero #digital #photo #photoframe #piframe #project #diy #foss
Oh wait, it's even more stupid. It sends the whole content of the textbox after every keystroke. So with each letter typed the request payload grows larger and larger. So with longer comments or posts EACH REQUEST is like 200-300 B. It sends 200 B every keystroke. So with average 40 WMP it's 200 letters per minute, so it's 40 kB/minute from every user.
I may be stupid but...
Can someone explain to me, what's the reasoning behind Reddit sending EVERY KEYSTROKE in a comment as a separate API POST request, instead of waiting until you clikc "Send" and send the whole text at once?
This approach results in 200 ms input lag when typing, which is super annoying and distracting
@sync_in however, I do wonder - what's the story behind this project? The initial commit was in July 2025, whit basically the whole, finished application dumped onto Github, but the company was founded in 2012.
I doubt the founder was twiddling his thumbs for 14 years before he actually got around to writing Sync-In...? So what's the deal about it?
So I've been looking for an alternative for #Google #Drive to host on my server. Very minimal requirements - only file storage, #WebDAV support, #Collabora integration (need mobile editor, and not going to pay OnlyOffice a single penny).
#Filebrowser #Quantum was closest, but (for now) it only supports OnlyOffice. Was also impressed by #GroupOffice, but the UI is very clunky and did not get "wife approval mark"
Finally landed on @sync_in. Fits all my needs perfectly!
@Gina hi there!
Afaik, the toot character limit on Mastodon is set on the instance level. Would it be possible to increase this limit on Fosstodon?
It's kinda annoying, trying to fit a meaningfull and elaborated discussion in 500 characters, especially since this has to also account for handles, hashtags and trigger warnings. As a result with longer threads you're often left with like 200 characters for the actual toot 😅
For example now I have only 24 characters left
But don't boycott a fucking local database for adding an optional column. I know people desperately want to be offended about it and hate systemd, but at this point it's like your local coffee shop added to the menu option for lactose-free milk, and you're angry at a piece of paper that the menu was printed on.
If the coffee shop decides that lf milk is the only milk you can order, then be angry at the coffee shop, not the piece of paper.
No. It did not. The only thing that was added was an OPTIONAL, EMPTY BY DEFAULT field in the userDB. To accompany fields called "Name", "Surname" and "Location". Want to fill it? Fill it. Don't want to fill it? Don't fill it. Or lie about it. Fuck cares. It's an OPTIONAL COLUMN IN YOUR LOCAL DATABASE. NOTHING MORE.
If or when distros start requiring you to fill your DoB - go ahead, be offended, go with pitchforks and boycott the distros.
This is the best evidence, that Google is going too far.
I got email, that my 200 GB Google One plan is being phased out (with wife we have shared sub for Google Photos), and I have to choose between 100 GB (too small) or 2 TB (too expensive).
Wife, who not even 2 weeks ago asked me how to open file browser on her MacBook, asked me if I can make "some server thingy", to have our photos locally, and not in Google Photos.
So here I am, configuring Immich locally!
Hmm I'm at a crossroads...
I self-host my Matrix instance, have some encrypted rooms there, chat with people from other instances. Works fine.
But Hetzner just raised the prices, and my backups are getting bigger and bigger. It's now 8 Euro / month, and if the backups keep growing at this rate, by the end of the year it could be 10-12 Euro / month.
What do I do... Suck it up, or just switch to public instance...
Difficult choices...
