Re: commit: add g_free to code generated

[Frank Rehberger <Frank Rehberger web de>]

Title:

Michael Meeks wrote:
> Hi Frank,
>
> On Sat, 2003-11-15 at 22:40, Frank Rehberger wrote:
> > The patch affects generated code by orbit-idl compiler.
> >
> > Generated destructors will free allocacted memory finally, ie:
> >   
>
> 	This is weird, I see a patch like this going constantly to and fro in
> this area. The change looks fine - but why are people using this
> -skelimpl mess ? also, what were the last few changes to this code
> about? I've lost track of what's up there.
The "-seklimpl mess" did generate buggy code but was shipped with
ORBit2: the impl_*_destroy was mixture of "deactivator" and
"destructor". This was wrong, as servant might be activated for
multiple POAs and, usually, deactivation is done independent  triggered
by user; finally destructor would be called from within POAManager.
 First patches reduced functionality to "destructor", for some reason
the final "g_free" has been lost, as noted in mailing list short time
ago. 

More or less, all this relates to "Beginners docs", writing samples for
CORBA C-mapping, easy to understand, correct, robust, etc.  Many
samples focus on handling of parameters, CORBA-sequences, exception
handling etc.  being usefull to more advanced bonobo-applications, too.
  Finding bugs they are fixed. Hope this helps?

Best regards, 
Frank