On Mon, Nov 12, 2007 at 10:11:45AM +0100, strk wrote: > On Mon, Nov 12, 2007 at 10:00:43AM +0100, Udo Giacomozzi wrote: > > Hello strk, > > > > Thursday, November 8, 2007, 7:30:01 PM, you wrote: > > s> Can we provide two different point_test_local, one for > > s> dynamic and one for static ? > > > > Rather I'd provide special path access methods for DynamicShape. Like > > a method to get the really defined paths and a additional method that > > closes the paths on-the-fly and returns it without modifying the > > definition itself. The latter would make happy both renderers and > > point test.. > > But would also mean a lot of memory copies. Closing on the usage side > would just be an additional final check. > What you ask is already available explicitly copying a path and calling > ::close() on it (in case it defines a fill side and is dynamic).
Another problem (which is present already, since we're closign the path) is that distance from the stroke should *not* be checked on the virtual (added for closing) edge. So point_test (and maybe the renderer too) should know what's a real edge and what's just a virtual one. --strk; _______________________________________________ Gnash-commit mailing list Gnash-commit@gnu.org http://lists.gnu.org/mailman/listinfo/gnash-commit
