[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kfm-devel
Subject:    Re: Bug#5114: Random crash
From:       Lars Knoll <Lars.Knoll () mpi-hd ! mpg ! de>
Date:       2000-06-19 10:44:28
[Download RAW message or body]

On Mon, 19 Jun 2000, Simon Hausmann wrote:

> 
> 
> On Mon, 19 Jun 2000, Lars Knoll wrote:
> 
> > On Sat, 17 Jun 2000, Waldo Bastian wrote:
> > 
> > > I have seen this QTimer appear more often in konqy bugreports... anyone an 
> > > idea which timer this is? Note also that we see the by now famous 
> > > DCOP-reentrancy in action.
> > 
> > There are only three QTimer's used in khtml. One in khtml_part for delayed
> > redirects, the second in khtml_part for autoscrolling, and the third one
> > in the javascript stuff. My guess would be the redirection timer. Some
> > sites (i vaguely remember cnn.com does that) set a redirect to itself with
> > a timeout of half an hour or one hour, to get automatic reloads of the
> > page. What if the part gets deleted before the timer timeouts?
> 
> Hmmmm, I think it should work. We're using a singleshot timer for the
> redirection, right? And AFAIK QSingleShotTimer internally uses a QSignal,
> which inherits QObject -> it should handle things properly if the document
> gets destroyed before the timeout happens -> the signal should go into
> nirvana :)
> 
> But what might happen is:
> You load cnn.com, the singleshot timer is started, some other page is
> loaded (without KHTMLPart being destructed) and then the timeout happens.
> This should successfully call the slot, but I guess the redirectURL value
> would be wrong or at least be empty.
> 
> Hmmm, I think it's safer to allocate a local redirection timer in
> KHTMLPart, instead of using a singleshot timer. Then we can stop it ASAP
> in ::openURL.

Yes. Sounds cleaner to me.

Lars

[prev in list] [next in list] [prev in thread] [next in thread]