[PyKDE] Weird problem with a QTabWidget subclass that overrides insertTab

[Phil Thompson]

I suspect that Qt is implementing addTab() by calling insertTab() - or vice 
versa - so when your re-implementations are getting called is probably more 
complex than you expect.

To confirm this you could rebuild PyQt with tracing enabled (and use 
sip.settracemask).

If you rename TabWidget.insertTab() to insertTab1() (and change the call in 
TabWidget.addTab()) then you get the behaviour you are expecting.

Phil

On Sunday 18 July 2004 6:31 am, Matthew Scott wrote:
> Hi all,
>
> I am diving head-first into PyQt again.  This will be the third time in
> as many years, and the third time I am enjoying the experience :-)  This
> is the first time I'm in a spot where I'm doing stuff that doesn't
> involve any QT Designer usage though, and the first time I'm subclassing
> built-in widgets extensively.
>
> My problem is that of overriding methods in subclasses of Qt widgets.
> Say you have a class Y which is a subclass of X and overrides the method
> z.  You want to call the z method from the X class as part of the
> overridden method.  The following code will print "abcdef":
>
> class X:
>    def z(self):
>      return 'abc'
> class Y(X):
>    def z(self):
>       return X.z(self) + 'def'
> print Y().z()
>
> So far, so elementary.
>
> Now we get to the tricky part.  You'd expect the following code to print
> this:
>
> addTab(somethingElse='bar')
>    calling self.insertTab
> insertTab(somethingElse='bar')
> insertTab done
>
> But it doesn't.  Instead it prints this:
>
> addTab(somethingElse='bar')
>    calling self.insertTab
> insertTab(somethingElse='bar')
> insertTab(somethingElse='foo')
> insertTab done
> insertTab done
>
> Here is the offending code, blank lines stripped out for brevity:
>
> import sys
> import qt
> class TabWidget(qt.QTabWidget):
>      def addTab(self, widget, label, somethingElse='foo'):
>          print 'addTab(somethingElse=%r)' % somethingElse
>          print '  calling self.insertTab'
>          self.insertTab(widget, label, somethingElse=somethingElse)
>      def insertTab(self, widget, label, index=-1, somethingElse='foo'):
>          print 'insertTab(somethingElse=%r)' % somethingElse
>          qt.QTabWidget.insertTab(self, widget, label, index)
>          print 'insertTab done'
> class MainWindow(qt.QMainWindow):
>      def __init__(self, parent=None, name=None):
>          qt.QMainWindow.__init__(self, parent, name)
>          self._setup()
>      def _setup(self):
>          self.tabWidget = TabWidget(self)
>          self.setCentralWidget(self.tabWidget)
>          self.someWidget = qt.QWidget(self.tabWidget)
>          self.tabWidget.addTab(self.someWidget, 'some widget',
>                                somethingElse='bar')
> class App(qt.QApplication):
>      def __init__(self):
>          qt.QApplication.__init__(self, sys.argv)
>          self._setup()
>      def _setup(self):
>          self.mainWindow = MainWindow()
>          self.setMainWidget(self.mainWindow)
>          self.mainWindow.show()
> if __name__ == '__main__':
>      app = App()
>      app.exec_loop()
>
> The problem is in TabWidget.insertTab, since it tries to call the
> superclass's insertTab method.  But it doesn't quite do that.  It
> instead ends up calling ITSELF again before it finally calls the
> superclass's method.  By doing so, it loses the custom 'somethingElse'
> that was set, which in my real-world code means that the overridden
> method becomes somewhat useless :)
>
> FYI, if you rename insertTab to insertTab2, and have addTab call that
> instead, and have insertTab2 still call superclass's insertTab method,
> problem "solved" and you get the expected output.
>
> But this seems weird, and a bit silly in my opinion.  I really want to
> have my QTabWidget subclass have the same API as QTabWidget itself, with
> just a few minor options added on.
>
> Has anyone else run into this problem who can offer a solution?
>
> Or perhaps, does anyone have a clear idea of why the above code is
> exhibiting this behavior and what a "better" way of doing it would be? :-)
>
>
> Thanks very much,
>
> - Matthew
>
>
> FYI - I am running Debian unstable with Python 2.3.4, and version 3.11-4
> of the python2.3-qt3 package installed.  I'd be curious to know if the
> code given above acts differently on other systems or versions of PyQt.
>
> _______________________________________________
> PyKDE mailing list    PyKDE at mats.imk.fraunhofer.de
> http://mats.imk.fraunhofer.de/mailman/listinfo/pykde