@hywan @Keltounet Question: "can the state of shells be improved enough to overcome the inertia of sticking to what you know?"

This is the wrong question. It presupposes zero cost of transition, while the cognitive workload of learning a new shell rises exponentially with age (hint: I'm nearly 60, shells are harder to adapt to than a new GUI). Stability and continuity are essential prerequisites to productivity!

@cstross @hywan we could say the same for languages, both in real life and computing. I'm 56 and enjoying learning both Rust and Japanese 😅

And looking into nutshell too.

@Keltounet @hywan Computing is not my job. It hasn't been my job for over two decades. Time spent learning a new shell or thinking about computers is time *wasted* from the non-compsci point of view.

Thing is, the question about the utility of switching to a new shell has embedded ideological assumptions that implicitly privilege computing over applications. To 99% of the world applications of computing are the priority; the machines and software are just an annoying drag on getting stuff done.

@tshepang @Keltounet @hywan UI changes in corporate products seem to happen purely to justify the product manager's request for a pay rise at their annual review! And at Google, old products get killed ruthlessly to make way for new products because that's the only way a manager can earn promotion.

This shit is actively hostile to the users' interests, but it's ubiquitous.

I mean, Canonical think 5 years is "long term stable" for Ubuntu, which is bullshit: should be AT LEAST a decade.

@cstross @tshepang @Keltounet @hywan I mean, on the other hand, 10 years LTS is... really expensive if not impossible. Computers are not some magical thing frozen in time. Maintenance costs rise fast, and not linearly with time, especially with context change.

I totally agree with your point on the UX change ofc, but for example, in terms of shell, as he mentioned in the article... Nushell is not really meant to replace your use of shell. It is built to replace glue scripts. And yes. needed.

@Di4na @cstross @tshepang @Keltounet @hywan the idea that a stable UI is something “magical” is pretty bizarre.

Required changes in gesture or keystroke should be carefully minimized and the vast majority should be optional.

Backwards compatibility is THE most important feature of any software product upgrade, but somehow that seems to be a forgotten principle. It’s perhaps only a small part of why we have a sense that our devices are losing rather than gaining utility. But it’s part of it.

@mtobis @cstross @tshepang @Keltounet @hywan ah yes, the most important part is to keep the UI that most of humanity cannot use.

Magical indeed.

@Di4na @cstross @tshepang @Keltounet @hywan

Nothing prevents a complete UI upgrade as long as the old one is still the default for existing users. But changing either the existing UI or any existing APIs for existing users is a betrayal. I don’t understand why this isn’t obvious.

@mtobis @cstross @tshepang @Keltounet @hywan and i don't understand how my point is not obvious either.

So maybe we should both realise our arguments are not as strong and evident as it looks.

@Di4na @cstross @tshepang @Keltounet @hywan

Yes perhaps I miss your point.

Mine is that i do fail to see why machines that are two orders of magnitude more powerful than thier predecessors can’t do exactly what their predecessors did in exactly the same way.

Maintaining software shouldn’t be a major expense. If it is, I think it is usually indicates a misdesign at some level of the stack. I can think of exceptions, but usually.

@Di4na @cstross @tshepang @Keltounet @hywan

Just fork it.

Geez. Is that so hard?

@Di4na @cstross @tshepang @Keltounet @hywan

I guess you're talking software services, while I'm talking local executables. But that's a key part of the enshittification process.

e.g. I just want to run the version of Photoshop I (thought I had) bought. I don't want Photoshop as a "service".