AW: Re: [omniORB] Tie mechanism and python

[andreas at roedl.ch]

Hi,

>On Friday 10 February, andreas at roedl.ch wrote:
>
>> I'm currently trying to understand why there is no tie mechanism for
>> the python implementation of omniorb. I mean, is there a reason, why I
>> can't create extra template classes for each interface with omniidl
>> -Wbtp -bpython <idl> ? Is there something comparable to the tie
>> mechanism for omniorbpy? Am I missing something?
>
>The Python language mapping doesn't specify a tie mechanism, so
>omniORBpy doesn't provide one. It only exists to make multiple
>inheritance situations simpler for C++ (and more significantly for
>Java). It's not needed for Python. Why do you want tie classes?

I want to tie my python classes - actually pyqt classes - to the skeleton classes of the interface. Every class in a pure pyqt application is inherited from a qt class, mainly from qobject or qwidget classes. I just started to use python for gui (qt) applications, but I'm almost certain, that I need tie classes to glue omniorb and pyqt together in a sane and elegant way.

Why do you say that the tie mechanism isn't needed for python?


Andreas