plasma memory/pixmap leak ?

[Aaron J. Seigo]

On June 2, 2010, R.F. Pels wrote:
> >> So, basically, the short answer is 'No'.
> > 
> > there is no instrumentation system, no. much like most other plugin
> > based software out there. if you'd like to develop something, that'd
> > be great.
> 
> Huh. Methinks that would require at least SOME community effort. Ehm,
> ISTM that this isn't something one can invent and develop on its own.
> Pointers? Where to start?

valgrind would be my suggestion for memory usage, though i'm not familiar 
enough with what valgrind offers to allow sorting out what plugin is 
allocating what.

pixmaps live in x.org and has only per-process reporting, but it could be 
possible to instrument QPixmap as Alexis already noted for there. there is 
Plasma::Svg which is the central place where all svg painting and usage 
happens, but that doesn't give us a full picture of pixmap usage.

i really don't have any good answers for this. doesn't mean they don't exist, 
of course.

> > we do have ways of finding out where problems are, though it's a bit
> > manual and involves running individual components. at least plasma
> > is highly componentized enough that things can be tested in relative
> > isolation.
> 
> Pity. Basically, we're forgetting to address the pesky issue of
> instrumenting our work 

there is a fair amount of instrumentation available, just not for resource 
consumption, which is not a trivial matter.

given that we have the ability to run one plugin at a time in things like 
plasmaengineexplorer and plasmoidviewer and tools like valgrind and 
kcachegrind for "big picture" views it's been a situation we've been able to 
at least live with.

-- 
Aaron J. Seigo
humru othro a kohnu se
GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA  EE75 D6B7 2EB1 A7F1 DB43

KDE core developer sponsored by Qt Development Frameworks
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://mail.kde.org/pipermail/plasma-devel/attachments/20100602/d4cc4a10/attachment.sig