[PyKDE] PyKDE update

[David Boddie]

On Mon, 29 Sep 2003 15:08:54, Gordon Tyler wrote:

> David Boddie wrote:

> > KHTMLPart *part = (KHTMLPart *) parent();
>
> I don't think that sort of concept really makes sense in Python.
> Type-casting in C/C++ just determines the particular interface through
> which your code is then accessing the object, it doesn't change the
> object's intrinsic type. Python only ever deals with an object's
> intrinsic type since it's a dynamically typed language.

Sure. The problem I think I have is that I'm always likely to be presented
with something that's wrapped as a QObject. I create a factory class based
on KLibFactory which is instantiated by a calling application. This creates
instances of a KPart.Plugin-based class on demand from the application;
its parent is a QObject.

Now, when my Plugin instance receives a signal, it checks its parent
widget/object using self.parent() only to discover that its a different
QObject. I suspect that this is because it gets embedded into a widget
by the application.

The point is that the parent is actually a KHTMLPart, but it could be
anything as far as I can tell.

> However, the problem may be that the sip translation is not translating
> that subclass. See my earlier patch for some subclass mapping that Jim
> extended and incorporated into one of the 3.8rc releases.

I'm using 3.8rc2, but I think that the above issue may lie outside the scope
of your patches. The subclass mapping works well in PyKDE; the above problem
is either the result of misinterpretation on my part, or something in a
grey area between C++ and Python. It's a case of working around the use of
pointers in the underlying framework which are only there to provide a
flavour of dynamic typing in a statically typed system.

As I understand it, anyway...

David