Transparent Ambition

Back in the 1990s I remember CS professors taking a very chauvinistic idea of what an "operating system" was limited to the kernel, would probably exclude a CLI interpreter like bash, never mind the rest of the userspace and would particularly exclude anything having to do with GUI even if you had kernel drivers and tons of DLLs that came with the OS to support GUIs.

Now you have Arstechnica, which should know better, which makes detailed reviews pixel-by-pixel of everything that changed between one version of MacOS and the next and seems to think the only thing that matters in an OS is the superficial things you can see, at least if the OS is MacOS.

Windows is refreshing because it has more widget sets than I can count but it doesn't matter because you can get your work done even though it is inconsistent and usually just a bit ugly. It beats Linux though, because at least in Windows if a label is 75px wide, Windows will make at least 75px of space for it, whereas in Linux nobody gets excited if it label gets clipped because they only left 55px -- they'll even close out a bug request about this as soon as you make it. But hey, Linux on a bad day looks better than the 99% percentile NFT.

This was sort of the idea of some of QNX. The kernel never crashed, but display, keyboard, networking, uart, etc was all user space. So yes you had a kernel that never crashed but frequently turned into a useless brick until rebooted

Yes, actual progress in meaningful operating system changes in terms of how one might write a program, or access fundamentals like identity, have really stagnated in terms of making their way down to end users.

We have high-concept things like Urbit that wanted to remake identity and secure communication that are utterly inaccessible to most savvy folks, never mind average end users. And in the meantime we're shipping entire Unix environments around inside OCI images to make up for the lack of consistency and portability on the backend.

Regular users are left with a tragedy of the commons as paid/saas programs support integrations with other specific paid products instead of with general open standards that could foster more diversity and interoperability. Everything is tacked on at the application layer - imagine going back to the folks who designed Kerberos decades ago and telling them that bouncing the browser back and forth between websites is what we settled on!

I like the new transparent theme, but yeah, it's just a GUI theme. Bring me back a consistent GUI where I can get themes from a modding community that apply to every app on the system and give me control over my look and feel! Instead, we get Wayland and a loss of 90% of the classic Linux desktop software. Bring me an OS that I can seamlessly deploy in a near-stateless/immutable fashion everywhere without falling into the trap of Nix! Instead, we get yet another rewrite of the fuckin' Ubuntu installer that still doesn't make the resulting system any more appealing for someone who would administer a corporate fleet of Linux workstations.

You've got to be a bit more specific when comparing 'Windows' to 'Linux'! Both operating systems are used with a dozen different graphics libraries, each with multiple possible configurations, and sometimes more than one for a single application.

The culture of various Linux-related communities also varies considerably. I guarantee you that the KDE community, for instance, is not going to stand for a label getting clipped. They are absolutely meticulous, especially where accessibility is concerned.

Maybe it should be /Linux/Red Hat/ as I've got personal experience with the GTK developers having a bad attitude about... everything.

One could make the case that theming in Linux is a problem, not a solution. You make your computer look like anything... as long as "anything" looks like crap.

I remember in the early 1990s theming was all the rage for Win '95 applications and the smaller the crapplet, the more aggressively it was themed. People got the sense that it was not cool and by the time Microsoft introduced really great theming support in WPF nobody wanted to use it.

The existence of so much software written to existing APIs is a reason why innovation in the OS seems at best marginal these days. If you thought "there is so much bloat in POSIX" you're either going to have to rewrite the whole userspace or build a POSIX emulation layer so you can use the GNU tools. If you wanted to build a really cool OS for the eZ80 it probably runs CP/M applications because if it did you have a lot of software to run on it and if it doesn't you have to write a text editor and everything else.

I had a situation in a car where the infotainment went on a crash loop. The infotainment. I was very glad the car stuff like odometer(which is also a display) and basic functions continued working.

Anyway my computer is not a car, nor my car is my laptop. The older i get the more i like the answer "it depends", because reality is like that, not pure or IFThisThenThat. It is a network of lots of knobs(weights) that together give a context dependent answer.