Hacker News Reader

Comment by AnIdiotOnTheNet | Hacker News Reader

I guess because I'm in a bit of a mood thanks to arguing with Linux Desktop evangelists lately, I'd like to point out that Rox Filer implemented exactly this file saving functionality (and many other good ideas RiscOS) on Linux. Then they built Rox Desktop around it. It was great. There were AppDirs, context-menu-only, spacial file management, drag-n-drop to save, etc.

Naturally, the project's reward for trying to improve the Linux Desktop experience was to have all its ideas ignored by the wider Linux community as it faded into obscurity.

If anybody was ever wondering why all these people who have gripes about the Linux Desktop experience never seem receptive to your pleas of "contribute! you can make it better!", this is why.

View on HN

I also gave up.

With KParts, DBus, Bonomo, GNU/Linux has the pieces in place to have something like Lisp Machines, XDE, Mesa/Cedar, ETHZ, instead replicating a PDP-11 is all that matters.

I make a point of not including any of the components that you mention in packages that I build.

Another comment by the author of the GP one states the problem, everyone who might work on this kind of thing has different preferences.

Faire, but then don't complain (as in the community) that macOS, Android, Windows and ChromeOS get all the nice desktop toys.

I've been thinking for a while that the only way to get a decent desktop experience on Linux is going to be to start from scratch. The graphics drivers and Wayland are probably adequate. But absolutely everything from there on up - window mangers, GUI toolkits, applications, configuration tools - needs to be written from scratch by people who know what they are doing and can build all of it as a cohesive whole.

This would be a huge project, and i don't know who has any incentive to do it. It's hard to imagine that it would be possible to make a business out of it.

I came to the same conclusion some time ago. There are of course things I find less than ideal about the Kernel, and Wayland, but overall they are adequate and would be a lot of work to replace. Pretty much everything else has to go though.

I've made some notes about how I'd build such a thing, but the truth is I have neither the time nor the skill set necessary to do all the work myself. Worse yet, when I talk to people about what I think a desktop should be like I often find that while we agree on some subset of ideas, we disagree significantly on others in irreconcilable ways. Either I haven't found the right community yet, or it doesn't exist.

There is Probonopd's (of AppImage fame) Hello[0], but unfortunately that also doesn't seem to have a lot of interest.

[0] https://github.com/probonopd/hello

Bit of an aside - but what potential do you think VR has for GUI's? Personally for desktop GUI's, I was happy with CDE affairs in the 90's, but then I'm a terminal baby so, a GUI was just a way to pull-up terminals for most of my needs, of which CDE still ticks that functionality well, albeit it don't have all the modern eye-candy, which I find a good thing distraction wise.

>The graphics drivers and Wayland are probably adequate. But absolutely everything from there on up - window mangers, GUI toolkits, applications, configuration tools - needs to be written from scratch

Doesn't projects like Enlightenment [1] do just that? They use Wayland and everything else(WM, Applications, libraries) was developed from scratch over it.

I had used it on Raspberry Pi 3, it was the smoothest desktop experience I've had on a Pi. But, unfortunately RPi wasn't officially supported and so none of the 3rd party softwares worked.

[1]https://www.enlightenment.org

I mean, this is what Android did, right? They use a Linux-based kernel but completely reimplemented the display systems. The real problem is that there isn't anybody who is willing to pay for that sort of work, either on the supply side or the demand one.

Not only that, Android IPC builds on top of Xerox PARC ideas.

Although it is very underutilized, you can compose Android applications (Activities) as Lego pieces, with each one doing just one thing right.

Please, no. One of the biggest problems IMO for Linux desktops is that everyone decides to rewrite most of it from scratch. (e.g. gnome 2 -> 3, Ubuntu's varied and changeable desktop UIs, etc...)

No-one can be bothered to fix bugs in the existing desktop environments because they are all too busy working on a rewrite that will be the new 'perfect' desktop, with its own set of annoying bugs and missing features compared to the previous revisions.

Linux needs desktops that get incrementally improved, fixing the countless minor but irritating 'paper cut' kind of bugs that hinder users. But the problem is, that kind of development work is boring, and the people who spend their free time writing open-source desktops naturally want to work on their own interesting projects instead. Reading and fixing existing code is just not fun.

This all sounds very harsh and unfair on the OSS developers, but it is just a fact of life; if you are volunteering your time and effort on a project, you probably don't want to do the boring and dull bug fixing work.

That is not entirely correct, f.i. there is http://trinitydesktop.org/ which continues to support KDE3 in current systems, respectively porting it to later versions of QT, fixing (security) bugs in the bitrotten external libraries, and so on.

Compared to contemporary DEs it flies!

The question remains if this makes sense, because countless other WMs can be made to fly also, which one could also use, and not lose that much functionality at all, given the lack of available software which would utilize anything that the old KDE3 has to offer. So one way or another, the Linux desktop experience remains ghettoish.

Are you thinking of something like NeXTSTEP ?

They did, it’s called android

You assume ROX was better than anything on the Linux desktop the time, that assumption would be false. Rox didn't do anything extraordinary or better than the desktops or window managers of the day.

> You assume ROX was better than anything on the Linux desktop the time

I don't assume, I know because I was there.

> that assumption would be false. Rox didn't do anything extraordinary or better than the desktops or window managers of the day.

Name a single modern Linux DE that supports drag-n-drop saving, or AppDirs.

An answer of "I don't use those features so therefore they aren't important" is not acceptable.

Reply on news.ycombinator.com