Version: (using KDE 4.2.0) OS: Linux Installed from: Ubuntu Packages The other day I noticed that my plasma process was using a lot of memory (the machine had slowed down, so I opened top to see if there was anything obvious). plasma did at that point show 370m in the RES column. At this point it was 2-3 days since I upgraded and started the computer with 4.2. I restarted everything, followed the progress in top and 10 hours later plasma was using 125m (as shown in the RES column). Since twitter was the only non-essential widget I was using, I shut it down, and now, a day later, plasma's RES is still 125m. I consider this "proof" for the twitter widget leaking fairly substantial amounts of memory.
I am also finding the same behaviour on the latest 4.2 builds from openSUSE. Plasma starts out using around 20Mb and a day or so later it's at over 100Mb. But only when I'm using the twitter plasmoid.
I can confirm the same behaviour, using openSUSE 11.1 with the FACTORY 4.2 packages on an hp 2133. With the twitter plasmoid on the desktop plasma memory usage grows from an initial 19 M to over 100 M at which point plasma becomes so slow it's unusable.
I am seeing this too. After running a couple of days, plasma hits 350+ megs of memory. After removing the microblogging widget and restarting plasma, plasma is coming in at ~50 megs.
Hmm...just tested twitter plasmoid with plasmoidviewer + valgrind and I didn't see any serious memory leak. Maybe it's a problem at some other place that gets worse when using twitter plasmoid inside plasma ?
https://bugs.kde.org/show_bug.cgi?id=184394 might be duplicate
Do any of you use analog clock, or has digital clock configured to show seconds?
I can confirm that memory usage of plasma is rising on my system, too. (OpenSUSE 11.0 with KDE 4.2 from Factory-Repo). But I don't use the twitter plasmoid. I have this behavior only from time to time. When I have it the seconds of my analog and binary clock start to jump irregular (bug 186013).
the twitter leaks have been fixed.
*** Bug 197263 has been marked as a duplicate of this bug. ***