KDE 4.3.0 Released: Caizen

Personally I never liked KDE a whole lot. But it's good to have for the folks that don't like Gnome. With the growing number of freedesktop standards that Gnome and KDE follow it's now possible to transition between the two without giving up much.

If KDE 4.3 truly matches were they were with 3.5 then that would be fantastic.

Uhh, as superstoned indicates, I think you're getting a bit ahead of
yourself there. I must admit that I'm still typing this from 4.2.4, which
I've finally been able to get working reasonably well, something that
simply wasn't possible with 4.1 let alone 4.0, but it took me about 80
hours of research, tweaks and workarounds to get it that way.

But I do think 4.x is moving faster than 3.x and have confidence that
KDE's moving in the right direction in general, or I'd have not stuck with
it that long, /especially/ that long! As SS pointed out, 3.5 was 3.5
years into 3.x, and we're 18 months into 4.x now. With six-month minors,
that should put 4.5 at 2.5 years in, and I DO think it'll be reaching 3.5
by then if not be 4.4. And even if it takes to 4.5 to reach it, that's
still 2.5 years to 3.5's 3.5 years, so still substantial speedup. And if
4.4 makes it, then that's only two years in, 18 months faster!

One of the big issues I have with 4.x is the lack of global multi-key
hotkeys (bug 161009, apparently no fix in site as the functionality is
apparently missing in a support lib now, qt4 maybe?). Probably 95% or
better of my app launches are hotkey based, and I HEAVILY used that in
3.x. Temporarily I've remapped to a meta-key setup, but it's not the
same. That's one of the workarounds, and given the bug status, I'd say it
can't have been fixed by 4.3 (tho there's a new app I found in Gentoo's
4.3 package sets, looking a couple days ago, that looks like it might be
related, intriguing me as to what might have changed...).

Another one that I REALLY hope is fixed in 4.3 is the infamous kde
immediately (konqueror and otherwise) timing out when trying to connect to
a local proxy (privoxy), as if it can't see it -- but firefox and kde3
have no problems seeing it. That one has had probably a score or more
bugs filed since 4.0, with several individual bugs fixed before 4.2.4, but
I still have the problem here, and I'm REALLY hoping 4.3 fixes it, as it's
REALLY irritating, particularly when plasmoids and the like that don't
have a shortcut to the proxy's own page (which apparently establishes the
connection and fixes the problem temporarily) exhibit the problem, since I
can then hit refresh repeatedly and not get anywhere, meaning that
(online, weather, etc) plasmoid is simply broken for that session. If
it's not, you can be sure I'll be filing the bug that I've been waiting to
file until I tested 4.3, since it was just about out. The good thing is
that I'm running Gentoo and can apply patches, etc for tracing and testing
if need be, so hopefully it'll go a bit better than for those running
binary distributions who don't know how to compile sources.

Another big one was a dual-horned bug, extremely slow compositing
reactions on a dual monitor setup too big to run OpenGL on my aging Radeon
9200, but which ran 3.5's compositing just fine on what is otherwise
exactly the same system both hardware and software. One horn on that was
that kde3 had a static desktop, and compositing a couple of static windows
and maybe a dynamic one on top of that wasn't too bad. But dynamically
updating plasmoids on the desktop trying to composite thru multiple layers
of windows simply didn't work so well! Once I discovered that and put all
the dynamic updaters on panels again, it helped a lot. The other horn was
that the default translucency delay was denominated in 100s of ms.
Apparently that's supposed to be total delay, but on my hardware, it seems
to be effectively per-step delay, so effects that should have taken half a
second were taking several seconds. After I figured that out and manually
tried tens of ms, now 1 ms, it works FAR better. Combine the two horns,
and I actually have usable response times now, with compositing enabled,
similar to what I had on 3.5.

That's a sample of the dozen or more issues I faced. No way can all of
them be fixed for 4.3, so it's certainly not yet approaching 3.5 here.
But the rate of improvement has indeed been very high, and 4.2.4 is
actually usable now, with the issues worked around or whatever. So I've
every reason to believe 4.5 will be comparable to 3.5, if 4.4 doesn't
reach it which it might, and indeed, that'll be a year to 18 months faster
than 3.x moved, so very good indeed, even if it's not quite to what you
were saying.