dce/rpc "client" api

[Elrond]

On Fri, Aug 18, 2000 at 12:11:12AM -0500, Peter Samuelson wrote:
[...]
> > I think the void* return value is the only way to generically pass
> > back information from the functions unless they all share a state
> > information for representing a connection.
> 
> void* is horrible for documentation and readability, unfortunately.  It 
> gives no clue as to what is supposed to be behind the pointer.  Yes,
> it's supposed to be opaque, but you could still typedef it:
> 
>   typedef void* rpcconnstate;

The other way is the one, that Gtk+ uses. It's a nice
thing. (At least I like it).

struct _GtkObject
{
	/* This is a pointer to the function table */
	GtkObjectClass *klass;

	/* other basic stuff */

	/* reference count.
	 * refer to the file docs/refcounting.txt on this issue.
	 */
	guint ref_count;
};
typedef struct _GtkObject GtkObject;

(next one modified to show stuff)

struct _GtkAdjustment
{
  GtkObject object;

  gfloat lower;
  gfloat upper;
  gfloat value;
  gfloat step_increment;
  gfloat page_increment;
  gfloat page_size;
};

struct _GtkAdjustmentClass
{
  GtkObjectClass parent_class;

  void (* changed)       (GtkAdjustment *adjustment);
  void (* value_changed) (GtkAdjustment *adjustment);
};


I think, this is one of the nicest ways of doing OO in C.
Gtk+ also has a lot of tools to do the Casts and verify the
casts are done in the right way (casting a GtkAdjustment to
a GtkObject is fine, the other way round gets you a big
warning)


So what I would propose:

There's only one public function for each transport, the
connect function, it returns a pointer to a struct, that
mainly contains a pointer to your function table
(cli_connect_fns).

You can then have a nice wrapper
void cli_connection_free(cli_rpc_info *con)
{
	if (con->cli_connect_fns->cli_connection_free)
		con->cli_connect_fns->cli_connection_free(con);

	DEBUG(0, ("..."));
}


Just another way of doing it.


    Elrond