On 5/19/06, Paul Davis <[EMAIL PROTECTED]> wrote:
I agree with this whole heartedly. If we start making assumptions about what gets passed to this class, it starts to lose some of its use. Unless theres a very very good cause for doing so I'd recommend making any assumptions.
I assume you mean *NOT* making any assumptions :)
Even though we might be thinking of using this to wrap tree models around STL containers, whats to stop someone from passing it a linked list of some non-standard format. Or what if we wanted to do something even simpler and just pass it an NxM dimensional array?
There's not really anything to stop you from passing in a non-standard container object, other than the fact that the compiler will complain if it doesn't define a begin() or end() method, or if it doesn't implement operator++ or something like that (just examples, I don't know yet exactly which methods will be used in the implementation). I've been focusing on the standard STL container implementation because I think that's probably the most common use case. But theoretically you could maybe have a couple different custom treemodels, e.g.: template <class T> ContainerTreeModel and template <class T, int N, int M> ArrayTreeModel or something like that (where N and M are the array dimensions). It doesn't seem like there would be much shared functionality between these two types, but I could be wrong. I haven't thought about it too carefully yet.
Looks good so far. And I sure am glad I'm not the one scratching my head about virtual inheritance and GObjects...as much fun as it sounds.
heh. Thanks for your input though. It's greatly appreciated. Jonner _______________________________________________ gtkmm-list mailing list gtkmm-list@gnome.org http://mail.gnome.org/mailman/listinfo/gtkmm-list
