Fix for style change crash, and code cleanup

[Dirk Mueller]

On Friday 03 October 2003 18:57, Maciej Stachowiak wrote:

> 1) operator++ still has to be changed to a method that takes an
> argument, because it depends on what used to be global statel. So the
> .increment( bidi ) would just turn into .increment( this ).

Well, you could fix that by letting the iterator store a pointer to the Bidi 
state object. I'm not sure if its worth it though, but it would produce nicer 
looking code. Maybe "next" is a better name than increment btw. 

> 2) several of the affected functions in bidi.cpp are already member
> functions of RenderBlock, so they can't also be member functions of
> BidiState. We'll need to either leave BidiState's fields public or else
> provide inline getters and setters.

Yes, nothing wrong with that. There are only few hooks into Render(Flow|Block)

> Hey, it didn't even take me 5 hours to code it, so I'd be surprised if
> it took 5 hours to apply. :-)

Did you ever read the diff between the Safari and the Konqueror version of 
bidi.cpp? Its like removing all the lines, and adding them all back in. 


-- 
> Looking for a KDE-related EMail-Alias ? Get one at kdemail.net for FREE! <