Also, should enabling/disabling menu items be handled in a similar way, using a callback, or should menu items be explicitly enabled/disabled by IO actions like enableMenuItem and disableMenuItem?
Preferably the latter. If a toolkit supports "tear-off" menus, the menu can remain on screen for an extended period, so the status of the individual menu items can change while the menu is displayed.
Relying upon an on-popup callback is likely to result in tear-off menus not working properly.
Well said. I also vote for the latter. Why? it is simpler, it can be implemented anywhere, it is well understood, it is needed anyway to support tear-off menus, and finally, a toolkit can optionally emit "initmenu" events for application writers that want to optimize their calls to enable/disableMenuItem. -- Daan.