If Linux and BSD had their shit together and had built layers to deal with services, storage and networking in a distributed and programmable way, maybe we would not be stuck with Kubernetes and shitty cloud services. The UNIX paradigm stopped evolving a long time ago.
@[email protected] plan9 could have saved us, but the world is cruel.

@rozenglass Given that the Plan9 clique lead to the Go language, I'm actually happy that it failed.

The Lisp machine model made so much sense, but history happened.

@[email protected] I would guess the surrounding community matters, Plan9 was being made at Bell Labs, Go at Google, the same people may have produced wildly different products given the different environments and different goals. I don't claim strong familiarity with either, but Plan9 seemed to be heading towards an interesting alternate tech-tree path. But yeah, I would choose Lisp machines over another Unix-like world any day. But I wonder, has there been any parallels in the Lisp machines world to the distributed OS resources paradigm of Plan9?

@[email protected]

I used to believe so.
So much that years ago I created #JehanneOS ( http://jehanne.h--k.it/ ) to cleanup the (few) #Plan9 issues I saw and make it both simpler and more poweful.

However after years working on it I gave up, mostly because I realized that the language of Plan 9 is just as arcane and elitarian as UNIX one.

Awk? Grep? Xargs?

Why do we use such arcane words?

Plan 9 is way simpler than modern unices such as #Linux or #BSD, but still too arcane for most people to learn.

#OberonOS was simpler, for example (but lacking important concepts such as hierarchical filesystem and proper multiprocessing).

We should start from scratch I'm afraid, learning from past errors and trying new appraches.

@[email protected]
Jehanne is moving!

Jehanne - Jehanne is moving!

Jehanne Operating System

@giacomo
I don't get the gripe. The Plan9 suggestion is equally valid for a hypothetical UNIX clone with "user-friendly" command names.
@rozenglass @galdor

I feel the biggest problem is the approach of volunteer projects, kernel, apps, GUI, whatever. People could keep the parts of their project modular, and view the project as a distribution of sorts. That would keep it feasible to try a distribution with, say, the Haiku desktop on top of Plan9. But Haiku needs great multi-threading in-kernel.