Re: UrShape definition Part II

[Cyrille Chepelov <chepelov calixo net>]

Le mer, jun 27, 2001, à 08:14:43 +0200, Andre Kloss a écrit:

> > OTOH, every UrShape (if UrShape are really seen as objects by the
> > core) should have at least an Object as their first member
> > (usually named obj if it's an Object, elem if it's an Element, and
> > so on).
> In my opinion it will rather be an element (post if you disagree).

It's really up to you. Basically, you have these ancestors:
	Object -- when nothing else fits 
	Connection -- objects which connect others by something resembling a
line (the actual position of the object depends on the where the end points
are connected to)
	Element -- rectangularoid object, whose position is independent of
others (indeed, that's what Custom Shapes use)

then, a bunch of other ancestors for particular behaviours (zig-zag,
orthogonal, beziergons, etc.)
	
> /* urshape.h : defining the generic UrShape */

looks good so far.

>   /* The miniDOM stuff */

Hmmm... basically, what you call miniDOM is a tree of lists, isn't it ?
How about
    UrShape *parent_shape;
    GList *self; /* this->self->data == this */
    GList *children; 
instead of 
>   UrShape *parent_shape;
>   UrShape*[] children_shapes;
>   UrShape *first_child_shape;
>   UrShape *last_child_shape;
>   UrShape *previous_shape;
>   UrShape *next_shape;

?

	-- Cyrille

-- 
Grumpf.