[Digikam-devel] [Bug 195583] New: First start-up migrating from 0.9.5 is _very_ slow
Richard Ash
richard at audacityteam.org
Sun Jun 7 17:55:12 BST 2009
https://bugs.kde.org/show_bug.cgi?id=195583
Summary: First start-up migrating from 0.9.5 is _very_ slow
Product: digikam
Version: 0.10.0
Platform: Gentoo Packages
OS/Version: Linux
Status: UNCONFIRMED
Severity: normal
Priority: NOR
Component: general
AssignedTo: digikam-devel at kde.org
ReportedBy: richard at audacityteam.org
Version: 0.10.0 (using KDE 4.2.4)
Compiler: gcc (GCC) 4.1.2 (Gentoo 4.1.2 p1.1)
OS: Linux
Installed from: Gentoo Packages
When starting digikam 0.10.0 for the first time after upgrading from 0.9.5, the
progress dialogue that appears whilst it migrates settings gets almost stuck in
the "Scanning images in individual albums" stage. In my case, the first few
(older) albums shoot past quickly with lots of disk IO, then it seems to stop
dead, using full CPU and no disk IO. After much watching and experimenting, I
discover that it always stops on the first album that contains JPEG-2000 images
rather than ordinary JPEG or PNG images. There are no messages sent to the
console, other than ones about mal-formed images/metadata (I have some files
affected by old exiv2 bugs). These messages do still occasionally pop up after
it is "frozen".
Investigation with lsof reveals that it is still making progress on examining
images, but has now slowed down to doing between 5 and 10 JPEG-2000 images
(sizes 6-12 Mpixel) per _minute_. If I am patient, then the scan does
eventually make it's way through my collection (18GB), but because the progress
indicator can take 10 minutes or more between updates, the program looks hung.
I think this is a usability problem, because most people won't wait all
afternoon for it to sort itself out (I'm still waiting...). This was never a
problem for 0.9.5, which started quickly on the same collection / hardware.
I can confirm that ImageMagick's identify command on the same images does take
about 10 seconds per image, which I presumed was because the entire image was
being decoded for the colour stats.
--
Configure bugmail: https://bugs.kde.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the Digikam-devel
mailing list