[cable] Re: cswig/python inheritance

[Brad King]

Kaben Nanlohy wrote:
> Kitware developers :
> 
> I hope that you enjoy your new offices.
> 
> ***
> 
> Regarding lack of inheritance information
> in wrappers generated by gccxml/cableswig :
> 
> In the the function void CableSwig::FindWrappedBases() defined
> in file Insight/Utilities/CableSwig/Executables/CableSwig.cxx,
> a call is made to bool CableSwig::FindClassGroup() in such a
> way that FindClassGroup() always returns false, and guarantees
> that no wrapped base classes will be identified. Here is a patch
> containing a fix.
> 
> 
> --- CableSwig.cxx.orig  Tue Feb 24 22:49:56 2004
> +++ CableSwig.cxx       Tue Feb 24 22:53:58 2004
> @@ -486,7 +486,14 @@
>    for( std::vector<cable::Class*>::iterator i = parents.begin();
>         i != parents.end(); ++i)
>      {
> -    if(this->FindClassGroup((*i)->GetQualifiedName().c_str()))
> +    // below, if CableSwig::FindClassGroup() is not passed a pointer to
> +    // std::string dummy_group then it will always return false, and
> +    // wrapped base classes will not be identified. dummy_group is
> +    // otherwise unused.
> +    //
> +    // k. nanlohy -- 2004 Feb 24
> +    std::string dummy_group;
> +    if(this->FindClassGroup((*i)->GetQualifiedName().c_str()), &dummy_group)
>        {
>        ::String* s=  NewStringf("%s", this->TemplateName((*i)->GetQualifiedName().c_str()).c_str());
>        Append(bases,s);

Excellent.  Thanks for finding this.  I've committed the fix with credit 
to you in the CVS log.

-Brad