Hacker News Reader

Comment by c0l0 | Hacker News Reader

I was a very happy KDE user during the 3.3 to 3.5 days. I tried sticking with the desktop after the 4.0 release (there were SO MANY changes and grand new ideas under the hood!), but lost patience soon after the 4.2 release - it just wasn't ready for day-to-day use; all the rough edges made me bleed every day. So I made the plunge to Xfce, and didn't look back for more than a decade.

Until some three months ago, when I had to set up a new, additional machine at home, and I decided to try the new (really: current :)) Plasma Desktop on Arch Linux for funsies. By now, all may Xfce machines, except the one at work, have been migrated to KDE. Now even my wife uses it on her personal desktop - she's especially fond of Okular, KDE Connect, and digiKam.

View on HN

KDE did everything they could to tell people that it wasn't ready for general until 4.3, but never figured out how to get people to listen. Distros insisted on installing it by default.

There was good reason for them to release 4.0 when they did but nobody has figured out how to communicate releases only to those the release is for (this last isn't about KDE, humans in general have the issue)

They should have called it 3.999 instead of 4.0, or "4.0 BETA", or "4.0 (for developers only!)" or anything like that. Anything other than giving it a new major version number without a disclaimer that it wasn't ready for prime-time. When you give something a new major version number, distro packagers will either update to it or they'll be subjected to endless questions from users asking them why the package is outdated.

> There was good reason for them to release 4.0

What were those reasons? To me, it really seems like they rushed it out because the release of Windows Vista and iOS made them feel like they had to release something to seem current and up to date.

The reason is that KDE is mostly developed by volunteers. Many of whom had branches that have been developed and ready for several years but not released. They were losing interest. KDE had to move the train to keep those volunteers around.

If you don't ship it and some point it's gonna die.

So use the beta version to integrate everything and make sure it all works before shipping it to users. That's the point of having a beta. But shipping what should have been a beta release to end users as though it were finished has ramifications in how your project is viewed and whether people keep using it.

There were many alpha and beta releases. The point was about finalizing API changes of kdelibs.

You can stabilize APIs in a beta, though? What am I missing?

Nothing, it was a mistake that some are attempting to explain away as everyone elses fault for misunderstanding.

I did the same thing as the other poster, went to 4.x with KDE then promptly left as I fundamentally disagreed with so many of the things they were doing. I ended up at i3 and haven't looked back. KDE is a lot better than it was back in those days, but i3 is so perfect for me I'll never be a full time KDE user again. Not because I hate them, but because they're not i3.

Back then there was just a single "KDE" release, instead of separate frameworks, desktop and applications that we have now. KDE 4.0 being released meant that the frameworks used to build applications were now stable and that initial ports of desktop and applications on top of those frameworks were now available. They made it clear that end users should wait with switching until the whole DE stabilizes.

That's just how their versioning scheme worked. The exact same thing happened with KDE 2.0 and KDE 3.0, but there was a different audience back then and 4.x took much longer to stabilize (as Qt3 -> Qt4 migration was much heaver that others, including Qt4 -> Qt5), so it became a bigger problem there.

It didn't happen with KDE 5.0 because there was no KDE 5.0 - now you have separate releases with separate versioning schemes of KDE Frameworks, Plasma Desktop and KDE Applications (and many apps were still based on KDELibs4 on initial release anyway and only gradually ported to KF5 afterwards).

The distros did a shitty job there; they should have seen that 4.0 wasn't ready (esp. when the KDE devs were even saying so), and continued to fully support KDE 3.x until 4.x was really ready for prime-time. Instead, the stupid distro maintainers completely dropped support for 3.x, forcing people to either use KDE4.x or to stick with an older distro version.

The whole point of a distro is to make intelligent choices about the software to include in the distribution and package it up into a coherent system for users to use. Even moreso for distros that claim to focus on stability, where sticking with older versions of packages is the norm.

I had the same experience but never moved back from Xfce.

Is there any particular reason you've now migrated back to KDE?

What I particularly like about Xfce is that it's not very bloated. KDE always seemed quite bloated to me.

Can you define "bloated"? RAM consumption? Disk space? All are a cheap commodity these days. I've been using KDE for the past 8 years on and off, and uninterruptedly in the last two years. It's fast, powerful and rich.

Yes, all of those things.

Sure, they are cheap, but if you ever try using Xfce (or similarly small DE) for a bit you will quickly realise just how slow other desktop environments (and Windows/OSX) are. Unnecessary animations is another one.

I've seen a similar argument defending "bloated" websites but then whenever a minimal website (e.g., a text-only site) is posted, people are astonished at how snappy it is.

That being said, I last used KDE over 10 years ago and computers have definitely become faster since then.

https://itvision.altervista.org/linux-desktop-environments-s... According to this test it uses approximately as much RAM as GNOME and 1GB more than XFCE. I guess on older hardware with 8GB RAM XFCE would make more sense, perhaps even for mobile devices where battery life is a scarce resource.

A NUC-sized mini PC with Q1'21 Intel Celeron ran all three DEs with no noticeable difference in speed. Perhaps there's a difference for hardcore usage but it simply isn't applicable to the vast majority of use cases. So yes computers have definitely become much faster since 10 years ago that a lot of these "bloat" really aren't noticeable anymore.

I assert that one exception to this is compositing.

Even if you're on a gaming rig with a really beefy GPU, you might not notice that DE compositing is on, but you will notice the difference once you turn it off.

To test this, try dragging a window around on a high refresh rate monitor on LXDE (since that DE has no composition by default, iirc) and then on GNOME, KDE, et al.

The overhead of composition extends to games as well, which is why I like KDE for being the only full-featured DE that lets you turn composition off.

I also found it to be slow on my (now 7 year old) laptop. Not everyone has the luxury of frequent upgrades. And I tend to prefer speed over bells and/or whistles.

KDE hasn't gotten more resource intensive on my laptop (turns 7 soon) since it was when I installed it in mid 2016. It probably uses a 200-300 more MB of ram compared to XFCE, but the number of tabs open in my web-browser dominates any other memory usage on my system.

The only thing that slowed my system down was the file indexer (Baloo); some people have said better now, but I never re-enabled it.

That. I disabled Baloo because it's cache was growing uncontrollably. And Firefox with many opened tabs uses most RAM anyway, in that sense I don't see how much it's the problem of KDE.

Did Baloo replace Nepomuk?

How does Baloo compare with locate?

> Did Baloo replace Nepomuk?

Yes, or more specifically it replaced nepomuk-kde (NEPOMUK[1] is/was a EU standard for representing semantic data via RDF)

> How does Baloo compare with locate?

It indexes the contents of any files with MIME types that it knows how to convert to text. AFAIK locate only indexes paths, not contents.

1: https://en.wikipedia.org/wiki/NEPOMUK_(software)

> I was a very happy KDE user during the 3.3 to 3.5 days [...] So I made the plunge to Xfce, and didn't look back for more than a decade.

My experience too. I liked KDE 3.5 so of course the devs had to fuck it up by chasing whatever is the user interface fashion of the day.

So I went with xfce which doesn't change much.

Maybe I will try out KDE again sometime.

May I ask what the biggest changes/ deal breakers were? It is difficult to understand the whys without knowing about the usage-pattern was broken. I'm genuinely curious.

IIRC KDE4 wasn't as much breaking existing usage patterns but pushing intrusive things like desktop widgets and generally trying to lock me into tightly integrated system. I remember it was really easy to create a desktop note by accident and somehow really difficult to delete one. KDE3.5, or Xfce for that matter, felt more like something that is good for launching apps, managing desktop and some generic utilities. The clumsy UI design didn't also help.

> May I ask what the biggest changes/ deal breakers were? It is difficult to understand the whys without knowing about the usage-pattern was broken. I'm genuinely curious.

KDE 4.0 was famously broken at release. It's not even missing features that was the issue, it was highly unstable and not ready for release.

I was in this group and the major thing that annoyed me was the reinvention of the Amarok music player, destroying my favorite music player ever overnight.

Now that you remind me, the Amarok update was terrible. For some time I used to run a self-built version of 1.4 and I see there is now a fork called Clementine and a few clone projects. For me streaming services have mostly replaced local music files.

KDE4 was pre-pre-pre-pre-release quality. It crashed all the time, it greatly increased hardware requirements, and they were trying to push the whole "semantic desktop" thing that was a complete failure and was basically nepomuk running at 100% to "index files" (that would never be found).

The desktop thing was one of my major issues.

They fucked up desktop icons. It'd be akin to purchasing a new walkman and the volume buttons don't immediately change volume up and down, instead they're more generalized buttons so you CAN change the volume up and down with them, just not by pushing them and nothing more.

Someone somewhere had an idea and an intent and I didn't give a shit and it represented a fundamental problem in their mental model of what they should be doing, so I left.

It's great that it eventually got salvaged but it should never have happened in the first place.

Trinity (forked from 3.5) is still around: https://trinitydesktop.org/

One can truly see that KDE 4 was the right move long term.

It wasn't chasing the design. The major change was upgrade to Qt4 and they rewrote almost the entire thing from scratch. Long term I think it was for the best.

Same here. I still miss the KDE 3 interface: a bit ugly but usable, quick and stable. Xfce and gnome UX have always felt limiting to me

Reply on news.ycombinator.com