Sometimes, I wish other software products took compatibility as seriously as Microsoft does.

You need to clean slate the entire world because everything is interconnected. If , for example, you want files to be more than streams of bytes then you basically have to re-invent the Internet as well.

The older I get, the more I realize that we don't need a clean slate OS. If you pick any modern OS right now, the most annoying parts are never the parts that were from 30 years ago; it's almost always the newest thing.

The opposite is true. You can convert anything to anything at the boundary. The beauty of looser coupling

You can't convert anything to anything at the boundary and still have full interoperability. An example is resource forks on classic MacOS -- they never played nice with an Internet that assumes that files are only a flat stream of bytes.

I think that sounds like an issue of not converting something properly? The problem with converting is that it can be expensive... that's true.

I hear GNU Hurd will hit 1.0 any day now[1] :) /s

[1] https://news.ycombinator.com/item?id=37036851

Google's Fuchsia is still the best bet for a genuinely new OS that may at least find a niche.

Capability-based microkernels have been exciting research projects for decades, and Google has apparently made one that achieves acceptable performance, which was a significant challenge.

It would just grow quirks like that over time, just like every system before did.

It's impossible to make all the right decisions at the beginning any big project that never get changed.

The baggage of Unix is gold. The baggage of DOS is manure.

Like everything being text? Pictures? Structured data?

By the time it's mature, you'll wish for another one. That is, if you're even alive by the time that happens.

Haiku? (if finished)

To convey one's mood.

In seventeen syllables.

Is very diffic.

John Cooper Clarke

TempleOS may be the closest we'll get for the foreseeable future.