Hidden directories & sidebar treeview crashing -- Long but please read

[Joseph Wenninger]

Hi

I don't have that much time at the moment, but I looked a littlebit into
the problem. The problem is caused by the kdirlister, which was only
thought for containing information about one directory, but not for
whole trees. That changed, but there are still some problems from the
old times around. eg. has a global setShowingDotFiles, instead of a per
directory flag. Which causes the removing of all dot files if you move
from one directory with viewing hidden files enabled to one with viewing
hidden files disabled (INCLUDING the item, which is currently in
use->CRASH)

There are a few ways to work around that problem, which all have some
kind of drawbacks.

1) one dirlister / directory -> quite a lot class instances for nothing
2) don't use KDirLister -> some own (tree aware) implementation needed
3) set the keep flag of openURL to false -> only the current directory
is watched for updates (that's what konqy's treeview/listview seems to
do
4) don't obey the directory view modes of konqy's mainview and make them
configurable for whole directory trees

At the moment I'm trying to implement 4, since it is the easiest thing
to implement, but it has the disadvantage if you want to hide many
hidden directories eg in your $HOME directory, but not in other trees 


Kind regards
Joseph Wenninger

Am Sam, 2003-02-22 um 17.53 schrieb Stephan Johach:
> Am Freitag, 21. Februar 2003 20:12 schrieb David Faure:
> > On Friday 21 February 2003 17:30, Stephan Johach wrote:
> > > Hi Scott!
> > >
> > >  I guess this is exactly the same topic I reported in
> > >
> > > <200301262010.11253.lucardus at onlinehome.de>
> > >
> > > Unfortunately nobody replied. :(
> >
> > There's nothing more to say about it until someone takes the time to
> > actually debug the problem. Using valgrind would be a start, even for
> > someone who doesn't know the code...
> 
> David, no reply looks like "ignore this stupid translator guy". I already 
> stated that I am willing to recompile with debug info to provide a backtrace.
> If this is undesired then I am sorry for bothering the list.
> 
> BTW Valgrind did not even compile here, at least 1.9.3. The last time I had a 
> functional valgrind, it took more than 10 minutes to start valgrind running 
> knode on my machine, after some time I had to kill the process ...
> 
> Stephan
> 
> -- 
> http://www.wormsalt.de