Re: libgda++

[Chris Wiegand <cdwiegand home com>]

*PLEASE* replace the C++ bindings. Originally I thought that I knew enough 
C++ to make decent wrappers, however, I haven't had the time or knowledge 
to really make them good enough.

Sadly, I have recently gotten a new job, which has taken most of my time 
the past 2 months. Although I would like to continue maintaining the MySQL 
provider, and I plan to go back through that code to re-acquaint myself 
with it, I simply don't have the know-how or time to really do a good C++ 
implementation - although I *am* willing to help write test code for one.

At 02:57 AM 10/23/00, rodrigo linuxave net wrote:
> Laurent Marzullo wrote:
>
> > Hello all,
> >
> > I've take a look at libgda c++ bindings and I've some comments:
> > I don't know what is your plan for this library, so sorry if
> > my comment already met some of your current work :-). If your not
> > really working on it for now, I propose to do all the stuff that
> > I write here myself.
> > I've currently only look at gdaConnection.
> >
>
> C++ bindings seem to be abandoned, since there's been some time since last 
> time I had
> news from Chris Wiegand, its author.
>
> So, as I don't know too much about those C++ things, and if there are no 
> objections from
> other people on the list, go on with it.
>
> >
> > 1) Using namespace.
> >         At least 1 namespace must be used, say GDA to hold all class
> >         from the three library (gda-common, gda-client, gda-server).
> >
> >         Say just one for now, gda-client.
> >
> >         ------------------------------------------------------------
> >         namespace GDA
> >         {
> >                 namespace Client
> >                 {
> >                         class Connection;
> >                         class Providers;
> >                         ...
> >                 }
> >         }
> >
> > 2) Using C++ features:
> >         I speak about GList* for gdaConnection::listProviders,
> >         gdaConnection::listDataSources, etc.
> >
> >         Perhaps, it will be more in the c++ way to do :
> >
> >         class Providers
> >         {
> >                 public:
> >                         class 
> providers_iterator        :       iterator_traits< ... >
> >                         {
> >                                 operator ++;
> >                                 operator *;
> >                                 // etc ...
> >                         };
> >
> >                 public
> >                         providers_iterator      begin();
> >                         providers_iterator      end();
> >                         ....
> >         };
> >
> >         same for DataSources.
> >
> >         In this way, all algorithm define in the libstdc++-vx will be 
> ok for
> >         Providers. (I speak about find, copy, sort, etc...)
> >
> > 3) Connection.
> >         As I don't really like dynamic allocation, I thought of an object
> >         connection like this:
> >
> >         namespace GDA
> >         {
> >                 namespace Client
> >                 {
> >                         class 
> Connection        :       public  Gda_Connection
> >                         {
> >                                 public:
> >                                         Connection( CORBA_ORB );
> >                         };
> >                 }
> >         }
> >
> >         In this way a structure of Gda_Connection will be in class 
> Connection.
> >         So I think of a function of the kind :
> >
> >         // I saw that gda_connection_init alread exist ...
> >         void    gda_connection_init( Gda_Connection* , CORBA_ORB );
> >
> >         Wich will do exactly the same as gda_connection_new without 
> allocating
> >         memory.
> >
>
> well, here you've got a problem. Firstly, gda_connection_init is private in
> gda-connection.c, because it should not be called outside of it. And 
> second, you really
> need to call gda_connection_new, because Gda_Connection is a GtkObject, so 
> you really
> need to do so to initialize GTK signals etc. So, I think you'll really need a
> Gda_Connection *cnc member in your Connection class.
>
> Other solution for not doing it this way would be to make your own GTK 
> object based on
> Gda_Connection, but I guess you'll have to do something similar as I said 
> before, so I
> don't think it's worthy
>
> >
> >         and then, the constructor will be:
> >
> >         GDA::Client::Connector::Connector( CORBA_ORB p_orb )
> >         {
> >                 gda_connection_init( this , p_orb );
> >         }
> >
> >         And then you will use it like :
> >         --------------------------------
> >         CORBA_ORB                       l_orb;
> >         CORBA_Environment       l_env;
> >
> >         CORBA_exception_init( &l_env );
> >         orb = gnome_CORBA_init( .... );
> >         CORBA_exception_free( &l_env );
> >
> >         GDA::Client::Connection l_conn( orb );
> >         l_conn.Open( dbname , username , password );
> >
> >         // And then for a record set :
> >         GDA::Client::RecordSet  l_rs( l_conn );
> >
> >         GDA::Client::RecordSet::recordset_iterator      itr;
> >
> >         // print them on cout or on any other container (perhaps
> >         // any GRID widget ... )
> >         copy( l_rs.begin( "SELECT * FROM testtable" ) , l_rs.end() ,
> >                         ostream_iterator<string>( cout , "\n" );
> >
> >         What do you think of this ? (I repeat, I could do it or help 
> doing it)
> >
>
> please do it! As I said, C++ bindings are unmaintained.
>
> cheers
>
>
> _______________________________________________
> gnome-db-list mailing list
> gnome-db-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-db-list