Wait, I have a 7800x3D and 7900 XTX and feel like I’m getting exactly the performance I’d expect for 1440p gaming. What do I need to look into to see if I’m leaving performance on the table? I’m using Arch so latest rolling kernel drivers seem to be working fine based on my monitoring of card stats and “feel” when playing modern games. Since performance has been fine out of the box, I never suspected I could be missing something so it would be nice to verify one way or another.

CLI, nvimdiff 90% of the time. If I’m on a windows workstation, I might end up using git extensions GUI as it helps me visualize what’s happening a little better sometimes.

Been waiting for tree structure! Thank you for the hard work on this, love this project.

Hell, a determined person could even run two kernels at once.

Just add your own context menu shortcut for .deb files that runs sudo deb -i $_

Tab unload is neat I guess but it would be nice if it also prevented the tab from loading until you go to it. I often open a bunch of YouTube links in a row, to catch up on my daily channels. So I have 4-8gb of useless browser tabs sitting waiting for me to get to them because I started on my subscriptions page and middle-clicked the last day’s or so worth of videos. It’s not much more work to finish that step and add an extra step to unload all those newly opened tabs, but it would be a nice option to not load new tabs until they’re activated since they already open in the background.

Maybe there’s a config for it somewhere.

Yes mathematically it scales perfectly. Except in the end it still gets rasterized. Because you have carrying line thicknesses in the icons themselves, it’s important to check how the icons look at common scales / sizes on a regular monitor resolutions such as 1080p, 1440p, and 4k.

No no, I understand what it shows. I’m suggesting the blended version looks nice and could be an extra option.

Could be useful to check scaling. How’s it look really small to really large. Might help inform some of the line work decisions. Looks nice with all the colors!

Also, the way you present the bottom row with a blended gradient, might be its own nice set, preserving the blending so each icon has a bit of multiple colors to it.

I would make the case for proxmox on the machine so you can divvy up the hardware as you see fit— but also setup the hard drives as a zfs1 pool (1 redundancy failure allowed). This way you can make multiple isolated machines or use LXC containers directly for apps, services, etc. while benefiting from ZFS’s excellent performance and reliability. I would say that TrueNAS Scale has been a bit of a letdown for me because it feels bloated, easy to make mistakes with complicated setups, and I have less control over the hardware. I don’t like how updates have fully broken apps. That said it is a reliable ZFS wrapper with more bells and whistles in the UI over what proxmox offers— caveat being that both can do everything if you want to take the time to learn ZFS commands.

There is also the TrueNAS based alternative HexOS that is more beginner friendly for just getting a nice NAS setup fast while still supporting apps / containers.

Thank you! This is a wonderful post, I will take another shot this weekend and hopefully something will stick this time :)

Any advice? I’m trying to get a handle on it but I’m having trouble remembering anything or finding what to do in the first place.

Jetbrains Rider for C# and VSCodium for arduino / microcontroller programming.

I’m trying to learn my way around the tmux + neovim life but the learning curve might be too much for me.

Just a reminder that as long as you don’t need any kind of platform hosting or complex multi-user setup, git itself works fine on a remote machine as your server, even just on LAN. (As always, just setup an ssh key on the two machines so ssh commands are secure and don’t require passwords all the time)

> cd /my/repos
> ssh user@10.x.y.z ‘mkdir /home/user/repos/new_repo.git && cd $_ && git init --bare’
> git clone user@10.x.y.z:/home/user/repos/new_repo.git

What about a hard drive made of network pings?

https://m.youtube.com/watch?v=JcJSW7Rprio

I will always recommend Ben Eater’s breadboard computer 6502 project for anyone who wants to know how it works. The 8-bit breadboard computer project as the next step too, to really dive into all the pieces. But the 6502 project is a nice entry point into hardware itself as well as the basic components of processor and memory. How and what the 1s and 0s are doing and how to make them do what you want them to do. Getting up to a working character display and serial input for a keyboard to type is such a satisfying process that takes only a few hours if you kinda know what you’re doing and a few days if you know nothing.

I love Actual. It’s fantastic and easy to use. I use off-budget accounts and weekly / monthly reconciliation just to keep the general value of these accounts at stable intervals.

I have a slight bone to pick with the PWA version of the site though. After a couple months of using the PWA front end to keep my budget and transactions accurate manually, I opened the site on my desktop browser and it completely lost all that work due to a sync issue. Apparently the PWA for weeks had not remained in sync and so all manual entries were not making back to the server. But the app works so well I never noticed because it kept just working. Supposedly there’s an alert saying it’s not synced with the server but it’s not prominent enough. So if you use that feature (the PWA) then be sure it’s syncing often.

This is really a problem of human vs computer thinking.

F and f are two different characters, encoded differently. Ergo, File and file are different by raw bytes.

Some developers wish to make the interactions for the user more consistent and thus a case-insensitive filesystem is born. The problem is that this is such a low level place to make this decision.

A filesystem, as in the kernel level interactions for files, should be case-sensitive in that every character is a unique series of bits. But there’s nothing stopping a higher level api from helping users out. It would be sensible to have a case-insensitive desktop environment.

The low level functionality should remain intentional though.

As an engineer, I’m not looking forward to the entire generation(s?) of vibe coders who couldn’t explain what a byte is and the ways one might be stored on a system.

I still think I’ll stick with nala as my apt front-end but hopefully this will be a more robust backend.