KDE 2.2.2
loopkin writes: "Seems that the last KDE 2 is out. KDE 2.2.2 is faster and more stable and secure than 2.2.1, as stated in the Changelog. You will appreciate the trick that makes the icons load 5% faster in particular. Announcement is here. Please use mirrors for download, but original FTP is here.
Note as well that maybe for the first time, there are _official_ RH packages for a _stable_ release (7.2)."
I am now using WindowMaker too and seeing it up and running in 3 seconds (including the numerous applets I use) is really damn satisfying.
There are many good ideas behind KDE, for example it has been the best one when it came to deal accurately with furious trackpad moves while scratching over MP*s.
But I reckon it doesn't fit on a laptop which is supposed to be switched on and off quite often, hence losing some precious productive time waiting for a GUI to be up and ready.
I know I may not have understodd with question but just consider that KDE may also be problematic on "recent" hardware.
Trolling using another account since 2005.
The crossov rplugin has nothing t all to do with KDe... it's a Netscape/Mozilla plugin. It does work in Konqueror, but the KDe team had nothing to do with it.
You're thinking about reaktivate, which is the KPart in KDE-CVS which does essentially the same as the crossove plugin (runs windows AcitveX controls ), but with one big difference - its free, as in beer and speech. It's nowhere near ready for primetime yet though (I don't even think its planned for release with KDE 3.
"So it trades stability for speed."
Don't you? Otherwise you are saying that objprelink makes KDE slower but more stable.
My general rule of thumb ist that a speedup below 30% for GUI applications isn't noticed by the user.
Did anyone try KDE 2.2.2, yet ?
I haven't found a description of the trick to appreciate yet - I was expecting some cunning algorithm or something that would be interesting to read about...
Hi bero, I'm having a play with prelink, but having a few troubles - is there a web site or FAQ for it anywhere, I'm sure others will be having similar troubles
/lib/i686/libc.so.6: Could not parse `lookup 0x40000000 0x00007114 -> 0x40000000 0x00132b14 /0 _nl_current_LC_CTYPE')
(it cant prelink some stuff.../usr/sbin/prelink:
Thanks
kio-smb: don't leave smbclients using 100% cpu hanging around.
This has been really annoying me. I'm the sole Linux user in an office full of Windows 2000 boxes, and it's been pretty tough to evangelise Linux's interoperability with Windows while I have to keep killing zombie smbclient processes any time I use SMB.
I haven't had a chance to download it yet (deadline tomorrow, y'see) but this, along with the other speedups and so on, could finally mean it's feasible to start winning people over to KDE.
Good work KDE fellas. You are all very lovely indeed.
--
Karma: Chameleon (you come and go)
Say no to software patents.
On a p3 550 system I built at work from the ground up I was suprised that I did not get written up for computer abuse because I had it booting, via LILO, Slackware 7.X, Redhat 7.X, Win98se, Win2K and even BeOs.
I was curious about the speed of a default Slack and Redhat install and while not scientific, it was very interesting, indeed.
If there was ever a reason not to use static libs (a la RH) this would be one point to hammer home.
I had KDE 2.X installed seperatly on both boxes (yes, I know it is "wasteful" of space, humor me) and proceeded to get some benchmark utilities off of freshmeat.net.
You see, what I had noticed was KDE 2.X was "snappy" on Slack and slightly "dogged" on Redhat... so it set me to wondering if it was just the RPM install vs compile on Slack.
Turned out that was part of the problem/question.
Memory performance was about +/- 10% with in each other, but hard drive performace was the "killer" of KDE's performance on RH.
This is what I found using hdparm (plus switches that escape me at this time) turned on/off between SL/RH:
MB/s on the same ATA66 drive and even another ATA66 drive just to be sure.
No hdparm init: RH=3.6Mbs, slack=8.6MB/s
hdparm init: RH=8.4MB/s, slack=8.9MB/s.
Hummm...I says. With hdparm init'ed on RH, KDE was quite snappy, despite the rare stumble and thrash of the drive.
Oh, and a word of warning aboud using hdparm (also in the readme) on older drives: not recommended unless it can do > PIO mode 2, IIRC.
So, yes, HD speed does affect KDE more than you would think. Something to be aware of.
If it is not on fire, it is a software problem.
Let me tell you a few of the features in KDE that make me vastly more productive, and which I feel crippled without.
I spent many years using WMs such as CDE, Afterstep (1.0 is the only good version, IMO), WindowMaker, BlackBox, and so forth. I have also used GNOME quite a bit, as well as MacOS, various flavors of Windows, and so on. None of them made me want to give up my console (though in some cases I had to because I was doing web design or something). But with KDE, I don't miss the console at all.
Can't get excited ? Don't, that wasn't the purpose. I agree that the announcement here was a bit exxagerating about that change. But have a look at the icon loading code, and try to make it faster than that, then we'll talk.
How come everyone seems to think that developers make thing slow _on purpose_ ? When we find a way to make things faster, we do, even if the result is only a 5% difference. Small steps, but they accumulate. Would you prefer that we don't fix the things we find ?
David,
actually happy about his icon loading fix....
and disappointed everytime he reads Slashdot, by this habit of criticizing really _everything_.
PS: note that the announcement could have said "icon-loading speedup" and nothing else. You could at least appreciate that someone took the time to measure the actual speedup even if the result isn't huge.