[Kde-bindings] Splitting up the SMOKE library

[Richard Dale]

On Sunday 15 July 2007, Ashley Winters wrote:
> FYI, those hooks for splitting up smoke originally were added during
> this discussion:
>
> http://lists.kde.org/?l=kde-perl&m=102421278314207&w=1
Amazing - that was 5 years ago! The only reason that I didn't do this when I 
added the KDE classes to the smoke lib was that it was just too difficult for 
me to get the kalyptus parser and code generation working with the KDE 
classes, and also change the runtime to work with multiple smoke libs. 

Also I think I missed the above discussion as I was away raising money doing 
paid work at the time, and couldn't pay much attention to language bindings.

Another reason we need to break up the smoke libs is that the index into the 
methods table is a short, and we are getting near the limit of 32000 or so 
methods that will fit:

    typedef short Index;

We could just change that to an int of course, but I think is a good time to 
follow through Ashley's original idea, as we have plenty of time before KDE4, 
and it isn't so much of a problem for the existing qt-only bindings.

One of Ashley's other ideas was to base a Smoke v2 library on the Qt4 moc 
generated format for slots, but moc every class and every method in the api, 
instead of just QObjects and slots/signals. Then add the other features that 
smoke has to that such as virtual method callbacks for overrides etc.

-- Richard