GNOME’s memory footprint varies across Linux distributions, likely because each distro enables different background services by default.

well configured AI bots can avoid those instructions..

it's easier to change browser

404 - File or directory not found.

Thanks. Updated link to the home page. Didn't realize my SPA generated Urls don't work. Sigh.

i like that gpt-oss can be easly jailbroken, the issue is that the prompt needs to be sanitized before execution.. so nothing new

2025 is the year of Linux on the Desktop and the merit goes to M$

Finally the battery car scam is over

LLM-eval-simple a tool to evaluate LLMs on your prompts

https://github.com/grigio/llm-eval-simple

the blockch*in doesn't bite you

yarr is a fantastic selfhosted reader

This is what I'm using right now. I like that it has a built-in "reader mode" where it fetches the target article from the website and removes all the crud.

But I do have a wishlist of creature-comfort items that would probably never make it in:

* I go days/weeks without reading anything and trying to find out where I left off is a big pain. There doesn't seem to be a way to sort chonologicaly (only reverse).

* The only difference between read/unread items is a tiny gray dot in front of the article title. (I'd rather have the unread items stand out more from the read ones, with a different background, bold text, etc.)

* It would be nice to have a per-feed setting of whether to show the article as it appears in the RSS feed, or go fetch it from the web in reader mode.

Counterpoint, I've been using yarr almost daily for about a year and I can't say I share any of your wishlist items. I love how simple and elegant it is, and anything that makes the UI more complex or distracting would only take away from that.

I run it on a VPS so I can access it from phone+laptop and it looks great everywhere. I've only "augmented" it by throwing a basic rss bridge on the same server (well, the bridge is really single-file python script that generates rss feeds from other sources).