--enable-final broken

[Simon Hausmann]

On Sat, Sep 13, 2003 at 12:13:21PM +0200, Luciano Montanaro wrote:
> On Saturday 13 September 2003 06:08, david mattatall wrote:
> > On Friday 12 September 2003 19:45, Alexander Kellett wrote:
> > > On Fri, Sep 12, 2003 at 12:20:09PM +0200, Luciano Montanaro wrote:
> > > > Both kio/bookmarks/kbookmarkimporter_ie.cc
> > > > and kio/bookmarks/kbookmarkimporter_crash.cc
> > > > define
> > > >
> > > > static const int g_lineLimit
> > >
> > > fix it then
> > > sorry but i'm not supporting this crap
> > >
> > > Alex
> >
> > To what crap are you refering? Kbookmarks IE importer? or --enable final?
> >
> > Info on what --enable final actually does, is sparce. Why do you think it's
> > crap?
> 
> --enable-final is a hack, but it is useful.
> What it does is that it includes all the sources of a module in a common file, 
> which is compiled as one file instead of compiling all the files separately.
> It improves compilation time, at least for me, and reduces the object size 
> slightly. Maybe it will be obsoleted by some future version of gcc, but 
> for now I think it should be supported.
> 
> But maybe Alex refers to the Importer stuff, I see now it is deprecated...
> 
> In any case, here is my fix for the problem.
> It moves the constants to the class definition, where I feel that these 
> constants should really be.

It should be function local (as it's only used in those functions,
not anywhere else in the class) , and that's what's in the current
CVS now, too :)

Simon