It’s the default package manager in Bazzite and is once of the most functional packagr managers on atomic fedora.

Learning nixos had been a lot of fun for me.

Your comment unironically is something I prefer and one of my biggest pain points with Linux.

As a newb, I'm sure there's something called with a mycommonproblemd name that has a stateful interface. But sometimes that all adds up to make things feel complex. And it let's me make stupid mistakes, like I forgot to close or open a port on firewalld, or I disabled a container but forgot to commit a change to my systemd units.

It's nice to just have a nice file called myservice.nix that tracks the firewall port, name, systemd startup and update scripts.

I’ve always struggled with what should be in Claude.md that doesn’t belong in readme.md or a similar supporting file.

I tend to include a well documented justfile, so between the readme and that common commands are covered. If there’s a style guide it should be its own file, or summarized in the readme.

If Claude is making errors I tend to just update my global Claude file, but I haven’t updated it in 6 months — only to disable Claude signatures on generated commit messages.

README.md should be for humans CLAUDE.md for machines.

Most agents use README as a storage for EVERYTHING related to the project by default, which is annoying for humans who just want to figure out a) what the thing does b) how to install it. Then you start reading and there's some intricate documentation on how data flows through the application etc.

If you're only working on your personal projects with no collaborators, just a global claude file is just fine. Per-project files are more for things that are specific to that project.

That’s incredibly frustrating.

I’ve got a NixOS Qemu VM I use to run openclaw in. I had Claude help me set it up, and it runs local models on my own machine in a config based sandbox.

Why should Claude block or charge extra to work on that?

Why should Claude care if I have instructions for Hermes or OpenClaw in my project repos?

This fingerprinting is incredibly sloppy for how much access to a machine Claude code has.

> This fingerprinting is incredibly sloppy

What part of "vibe coding" is unclear to you?

These are the same people that use React as a TUI and render at 60FPS to your terminal in order to update a spinner.

Now you've learned the advantage of knowing how to do things yourself. When you depend on untrustworthy agents, you shackle yourself to their idiotic whims. Be careful who you partner with.

If it's just to set up a VM, how much would you even need to use? A couple of cents?

I run an OpenClaw VM and used Claude Code to build the VM scripts. The VM is connected to local llama.cpp, so OpenClaw and the models are running on my own physical hardware.

For what it's worth, the bars correspond in order with the legend. Plus there’s hover text.

That would be “permissive license”

Maybe we should have a little cue card for models: vendor/name, size, open weights, open source, permissive license.

It’s simple enough an idea.

I’m actually trying to move back from the Claude Code style, I feel like it’s easy to become distant from your own code, and I am feeling uncomfortable with that.

I’ve “vibe-coded” some projects and when I start to find issues or go to refactor them I don’t have that memory of why decisions were made, because many decisions were never made.

The early quants for Gemma4 26b had issues and needed to be updated, might be worth checking

It takes me 6 minutes minimum to get a response in the last 3 days, I don’t think model capacity is better.

It seems like they shifted heavily to prioritizing enterprise users. Starting in the last day or two I started getting much faster responses on an enterprise plan.

For what it’s worth, I tried making a GTK4 app. I got started, created a window, created a header bar, then went to add a url/path entry widget and everything fell apart.

Entry suggestions/completions are formally deprecated with no replacement since 2022. When I did get them working on the deprecated API there was an empty completion option that would segfault if clicked. The default behaviour didn’t hide completions on window unfocus, so my completions would hover over any other open window. There was seemingly no way to disambiguate tab vs enter events… it just sucked.

So after adding one widget I abandoned the project. It felt like the early releases of SwiftUI where you could add a list view but then would run into weird issues as soon as you tried adding stuff to it.

Similarly trying to build an app for macOS practically depends on Swift by Sundell Hacking with Swift or others to make up for Apple’s lack of documentation in many areas. For years stuff like NSColor vs Color and similar API boundaries added friction, and the native macOS SwiftUI components just never felt normal while I tried making apps.

As heavy as web libraries and electron are, at least work mostly out of the box.

There are definitely a shortage of good GUI toolkits - making one is a huge undertaking. GTK is mediocre, as you discovered.

QtWidgets is extremely good though, even if it is effectively in maintenance mode.

Avalonia also seems good too though I haven't used it myself.

I've used Qt off and on, and it's generally worked as advertised. Although when drawing very short lines on a canvas way back when (~2004), it wouldn't do a great job and I had to hack in custom routines that did a much better job.

For prototyping / one-offs, I've always enjoyed working in Tcl/Itcl and Tk/Itk - object oriented Tcl with a decent set of widgets. It's not going to set the world on fire, but it's pretty portable (should mostly work on every platform with minor changes), has a way to package up standalone executables, can ship many-files-as-one with an internal filesystem, etc..

Of course, I spent ~15 years in EDA, so it's much more comfortable than for most people, but it can easily be integrated into C/C++ as well with SWIG, etc.

> I've always enjoyed working in Tcl/Itcl and Tk/Itk

In the near future I need to lash up a windows utility to generate a bunch of PDF files from a CSV (in concert with GhostScript), with specific filenames. I was trying to figure out the best approach and hadn't even considered Tcl and Tk - with Itcl you might have just given me a new rabbithole to explore! Thanks! (...I think!)

I hope it works out! It's amazing how far Tcl/Tk has come since I "had" to use it as a wrapper around an X11 window back on an SGI Irix, using Tcl scripting to interface to an OpenGL backend. I think that was like 7.3.x or something in 1994. And it was pretty cool back then already! The team around Tcl is small, but dedicated and brilliant, IMHO.

>Although when drawing very short lines on a canvas way back when (~2004), it wouldn't do a great job and I had to hack in custom routines that did a much better job.

QCanvas (or was it QGraphicsCanvas?) has long since been replace with QGraphicsScene, which is much more capable and doesn't suffer from pixelation issues.

Probably. We paid thousands / year for the developer seat in our startup, and in the end, it wasn't great. I did manage to make the Tcl/Tk event loop and the Qt event loop work together, so we could have Tk windows inside a Qt app!

There is a small business licence.

Qt is still under very active development. Although there seems to be a lot more emphasis on QML than the widgets side of things for some time.

Yes QtQuick/QML is under active development. QtWidgets is not really.