Re: The g_object_{set,get} class of functions seem strange

[Havoc Pennington <hp redhat com>]

Skip Montanaro <skip pobox com> writes:
> In trying to figure out how I can get a list of property names for an
> object, I came across the g_object_set and g_object_get functions and their
> _valist versions.  Their signatures seem rather bizarre to me.  The get
> functions are declared to return void while the set functions are declared
> to return a gpointer.  How does g_object_get return values to its caller
> (g_object_get_valist does something weird with its var_args argument)?  Why
> doesn't it just allocate an array of gpointers and return them?  Why does
> g_object_set need to return its first argument to its caller?
> 

The return value is a conspiracy by Tim to confuse people. ;-)

Well, it is to support some kind of bizarre coding style:

 g_object_set (object, "child_object",  
               g_object_set (object2, "foo", 10, NULL),
               NULL);

Basically my recommendation is to pretend it returns void.

g_object_get() works like this, say "foo" is an integer property:

 int val;
 g_object_get (object, "foo", &val, NULL);

To get a list of properties that exist for an object you'd use
something like g_object_class_list_properties().

Havoc