[Kopete-bugs] [Bug 164979] New: ridiculous brain damage in the way kopete reads (at least) config files
Manuel Amador (Rudd-O)
rudd-o at rudd-o.com
Thu Jun 26 03:12:50 CEST 2008
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
http://bugs.kde.org/show_bug.cgi?id=164979
Summary: ridiculous brain damage in the way kopete reads (at
least) config files
Product: kopete
Version: unspecified
Platform: Ubuntu Packages
OS/Version: Linux
Status: UNCONFIRMED
Severity: normal
Priority: NOR
Component: general
AssignedTo: kopete-bugs kde org
ReportedBy: rudd-o rudd-o com
Version: (using KDE 4.0.83)
Installed from: Ubuntu Packages
OS: Linux
What I am about to attach is a heavily compressed strace output file from a started-from-scratch, configured Kopete (the latest version that came together with KDE 4.1 Beta 2). The application never finished to start because I interrupted it after OVER FOUR MINUTES on a dual core machine running ZFS. Yes, ZFS is slow. That's not the point. Please note the following things:
grep -E '(stat|open)' kopetetrace.2697 | wc -l
122432
122000 stat/open calls on a program that hasn't even finished to start but it's in a permanent grey window state? Upon further inspection, you see it statting thousands of times several kde4globals files and ALL its component directories. What happened to plain old just open()? Is the hundreds of thousands of statting some sort of foreplay for an operation that you will know for certain failed or succeeded? And what's with the statting of container directories up to the root directory? And why is this application statting the SAME PATHS over and over again? Don't you see this would kill NFS performance as well? Can't you cache the kde4globals files until you get an async reconfig notification or something?
You know what, I'll just let you process the file and figure out this brain damage. I honestly don't know why no one has measured the amount of work your library calls do, but Konqueror, a clearly more complex application, takes a fraction of the time to start than Kopete does (it probably does have quite a lot of syscall braindamage as well, and I'm speculating on how long kopete takes to start under these conditions, because I got BORED of waiting for it to start).
More information about the Kopete-bugs
mailing list