[linux-elitists] "Why I Hate KDE 4"
Karsten M. Self
Tue Nov 4 14:36:48 PST 2008
on Sat, Nov 01, 2008 at 12:24:10PM +0200, Shlomi Fish (firstname.lastname@example.org) wrote:
> Hi all!
> I wrote a rant about why I hate KDE 4 on my blog:
Shlomi's taken some hits for this post. His isn't a particularly
enlightened rant, but I find the genre of stuff that sucks (or doesn't)
generally useful. Free software has benefitted greatly from criticism,
though specific and directed bug reports tend to be preferred.
Even so, reasoned statements of preferences (positive or negative) can
be very useful guidance. The "essentials" post series from dive into
mark, in particular, which have documented his multi-year migration to
Mac, back to Linux (Ubuntu), and now Debian, is information and
instructive, particularly to people who might be testing the Linux
waters. Pointing out the gems, or warning that here there be dragons,
is very useful.
Mark's just posted Essentials, 2008:
...and the series is (somewhat) collected here:
... though that misses
... which was is dalliance into Mac OS X territory.
Shlomi's rant might be better served if it didn't start with the
whingey and piquant "it looks like shit", perhaps tempering that with a
Nielsenian UseIT analysis or screenshot / screen element comparisons to
KDE3 or alternative desktops.
I don't generally run KDE as a desktop (I do launch into it
periodically), finding WindowMaker generally sucks less (there's an
essay-in-process on my reasons why, nutshell summary: stable, reliable,
configurable as I need it to be, some elegant and underappreciated
features, stays out of my face).
All that said, some comments about KDE4:
- There's a lot of obvious focus on visual appeal, to the detriment of
usability. Even ignoring functionality, superposition of
low-contrast elements, visual feedback, and other design elements
- A pretty obviously rushed or premature release, especially on
nominally stable end-user distro versions such as Debian's upcoming
Lenny or the just-released Ubunutu Intrepid Ibex. I have to applaud
at this point Ubuntu's decision to exclude KDE4 from its LTS release
earlier this year.
- Some really annoying bugs. I am getting random keybaord lockups
(see bugs 156852, 123274, and 142585 at http://bugs.kde.org/), and
Konqueror frequently failing to find klauncher (bug 164307). Fixed
upstream, apparently, 2008/10/13.
- Konqueror find dialog disappears on each match within a webpage, and
must be re-invoked. Very annoying. "Find as you type" does not
repeat finding on any of the usual suspects (<enter>, "/", "n").
OK, now the <ctrl>-F hotkey for search dialog isn't working either.
- Wildly misbehaving panel apps. I think it was one of the clocks
which overflowed the panel onto the desktop, causing other panel
apps (workspace switcher, battery monitor) to also blow up in size.
Not being much of a fan of panels in the first place, I like them
even less when they misbehave.
- Some poor namespace decisions in app renaming. 'kcontrol' is now
'systemsettings', which is pretty broad and ambiguous. Defined as
"ksystemsettings' or 'kdesystemsettings' would be much more sane.
- I still don't see my own most preferred feature: a pinnable window
list. Heck, pinnable window menus in general, a'la WindowMaker.
I haven't dug into other apps yet (I generally use Konq for browsing,
Korganizer in corp environments for mail/PIM, amarok for music, etc.),
but I'm a bit leary from my experience to date. Pity as KDE has in the
past been the thinking-man's desktop (see Linus's endorsement for
example), but seems to have lost focus or succumbed somewhat to
As usual, a suitably tuned Google search turns up a number of
discussions of this topic:
Rant though this is:
- I generally prefer KDE's approach and attitude to GNOME's.
- I don't run _either_ as a desktop most of the time. I try 'em from
time to time (mostly to remind myself of why I find them so
frustrating), but fundamentally appreciate the fact that I (and you)
have a choice.
- There's no real reason either might not come through eventually.
Though based on past experience my bets are on KDE.
This also speaks to the difficulties of GUIS and desktop development in
general, in open source development processes in particular, and the
best way in which adequate in-development testing of desktops can
What appears to happen is a quiet, insular development cycle, light
early adoption principally by strong advocates (a population which with
an appetite for novelty, flash, and visual chrome). Toward the end of
which cycle the product is presented to the larger user community, and
gnashing of teeth and complaints concerning stability, feature loss (the
feature you use is much more visible than the one you've never seen
before), various elemnt drift, and very often, resource demands
particularly on trailing-edge hardware commences.
I remember similar wailing around the time of the KDE 2.0 release cycle
(October, 2000). A large element of complaints then were about
the apparent from-scratch rewrite KDE (the release announcement claims
it was "almost completely engineered"). Early manifestations of this
included significant feature attrition and some spectacularly unstable
early code. I _don't_ recall the 3.0 release being as noisy (2002), and
suspect it was a more evolutionary project, with a significant part of
the effort being both enhancing existing and developing additional KDE
apps. I did find it almost usable (ran it for a month or so solid as a
work desktop, and frequently in its various Knoppix instanciations),
though still not sufficient to knock WindowMaker off its pedestal.
One of the problems, of course, is that stability and usability of your
GUI is key to any significant value of your computing platform. Testing
out early betas of a desktop or windowmanager is going to be an
inherently annoying process, and gets in the way of real work (though
this can be minimized by running additional or nested X sessions).
Another is the inherent problem of genererating, encouraging, and
incorporating useful bug reports and other suggestions from people who
very nearly by definition are NOT programmers. GNOME's attitude (if
you're a target user, you're not qualified to code; if you're qualified
to code, you're not a target user; if you're going to complain, submit a
patch) appears possibly flawed (some here may recall an earlier
discussion in which this emerged). I'm not aware that KDE has expressed
a similar mindset as an official or key-leadership opinion.
At the same time, I note that there are numerous window managers and
even a couple of other desktop environments which have followed a much
more gradual, evolutionary path. Though this doesn't lend itself to
wild new functionality, there's also the question of just what it is we
expect a desktop to do (a question I'd suggest the GNOME and KDE teams
ask themselves). Because GUIs don't generate the sort of associated
legacy infrastructure scripting and programming languages do, and
because there's a value given by some to novelty, I note that there is a
tendency to see a fair amount of change for its own sake (an affliction
present in both open source and proprietary software). Strip away some
GUI chrome and you'll often find the underlying scaffolding hasn't
changed much (indeed GNUstep's internals are apparently *very* similar
to Aqua from a few things I've read, the appearance is quite different).
Upshot is: there's an incentive for dramatic change, those who value
interface stability may find ourselves frustrated, and much of that
change is superficial. Is this really useful?
4.0 appears to have been another substantial rework. I suspect that the
effort will pay off, though as of 4.1.2 I don't believe it has. I would
strongly encourage KDE to be more mindful of namespace particularly on
generic-sounding tools. And to look at ways of improving the
development, test, and release process in ways which promote useful
innovations, preserve utility, enhance usability, and reward buy-in from
users, community members, and partners.
Karsten M. Self <email@example.com> http://linuxmafia.com/~karsten
Ceterum censeo, Caldera delenda est.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 191 bytes
Desc: Digital signature
Url : http://allium.zgp.org/pipermail/linux-elitists/attachments/20081104/b755a7ca/attachment.pgp
More information about the linux-elitists