From kde-commits  Mon Aug 22 09:00:47 2005
From: David Faure <faure () kde ! org>
Date: Mon, 22 Aug 2005 09:00:47 +0000
To: kde-commits
Subject: Re: branches/KDE/3.5/kdelibs/khtml
Message-Id: <200508221100.51433.faure () kde ! org>
X-MARC-Message: https://marc.info/?l=kde-commits&m=112470129204242

On Tuesday 16 August 2005 18:05, Andras Mantia wrote:
> On Tuesday 16 August 2005 19:36, David Faure wrote:
> > On Tuesday 16 August 2005 12:13, Andras Mantia wrote:
> > > > On Tuesday 16 August 2005 12.17, David Faure wrote:
> > > >> > SVN commit 449438 by kling:
> > > >> >
> > > >> > Make sure there's a defaultHTMLSettings() around when fetching
> > > >> > accessKeysEnabled setting. Should unbreak Quanta, as discussed
> > > >> > on kfm-devel.
> > > >>
> > > >> Shouldn't this new+delete only be done when
> > > >> defaultHTMLSettings() is 0?
> > > >
> > > > You are probably very right. :-)
> > >
> > > No, that would reintroduce the crash. defaultHTMLSettings() starts
> > > with an "assert(s_self);". The whole KHTMLFactory class is strange
> > > for me as it has static methods that cannot be called unless you
> > > create an instance.
> >
> > OK, try
> > KHTMLFactory::ref();
> > ... defaultHTMLSettings() ...
> > KHTMLFactory::deref();
> 
> ::ref() and ::ref() is private, so one should use a "friend" 
> declaration. I suggested on kfm-devel to not do it that way.

I meant inside the khtml code, not in quanta.

-- 
David Faure, faure@kde.org, sponsored by Trolltech to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice (http://www.koffice.org).