Isn’t xess capable of running on any GPU too?

That’s also true, but I have experienced an occasional issues when it would be stuck on downloading some package at 10 KiB/s because of bad mirror. Parallel downloads likely wouldn’t have helped in this case since it would select the same mirror. Obviously both issues need to be fixed though.

I’ve got a problem with port forwarding I can’t get working, never had that problem before and I don’t know network stuff well enough to figure it out.

Docs says that CachyOS has UFW firewall enabled by default. You can search how to configure it, it seems quite easy.

The updates are the winner for me- I don’t know how long this has been a thing with arch but downloading multiple packages at the same time. Game changer. I love Tumbleweed, but a 2gb “zypper dup” downloading package by package could take me 30 - 60 minutes.

It’s usually the issue with automatic mirror selection. If you interrupt zypper using ctrl-c (only when it’s downloading, not installing of course) then it should select a faster mirror next time you run it. Zypper devs really should work on this though.

I actually installed it recently out of curiosity, but I’m hesitant about learning its advanced features like that. At least jq is a standalone tool that’s more ubiquitous than nushell, so you can rely on it even in environments that you don’t fully control (e.g. CI like GitHub Actions). And if you use it in some public code/scripts then other people will be more familiar with it too.

Fish is a replacement of bash that’s a bit more user friendly (has some cool auto completion features out of the box and more sane behaviour like handling of spaces when expanding variables). I personally started to use nutshell recently but unlike fish it’s very different from bash.

Starship is a “prompt” for various shells (that bit of text in terminal before you enter the command that shows current user and directory in bash). I haven’t used it but AFAIK it has many features like showing current time, integration with git, etc.

I use it occasionally but every time I need to do something a tiny bit more complex than “extract field from an object” I have to spend half an hour studying its manual, at which point it’s faster to just write a Python script doing exactly what I need it to do.