[cairo] Re-working cairo_current_path

[Kristian Høgsberg]

Keith Packard wrote:
> Around 13 o'clock on Jan 24, =?UTF-8?B?S3Jpc3RpYW4gSMO4Z3NiZXJn?= wrote:
> 
> 
>>Ah, right, take the length out of the cairo_path_element_t and keep 
>>knowledge of element size inside cairo.  Yeah, that's better and makes 
>>it possible to adjust the path representation later on if necessary.
> 
> 
> If we need to change the path representation, we can create a new 
> function that returns the new type.  As long as this datatype is allocated 
> explicitly for the clients use and isn't simply returned as a pointer to 
> the internal state of the path, I don't see the benefit in making the 
> contents opaque.  It's not like applications are going to magically be 
> able to deal with a new component of the path...

Using an iterator and an opaque data type would make it feasible to 
actually use the cairo intenal representation for iterating the path:

	cairo_walk_path (cr, &iter);
	while (cairo_path_iter_next (&iter, &elt) {
		switch (elt->type) {
		case CAIRO_PATH_ELEMENT_MOVE_TO:
			...
		}
	}
	cairo_path_iter_fini (&iter);

... or something.

Kristian