CGAL users discussion list

[Ben Supnik <>]

Hi Efi,

Foo - I can't reproduce the problem.  My old code was based on building 
the map out of a series of boolean operations...because of the poor 
performance of this structure, I rewrote the code a few days ago to do a 
single massive insert_curves, and the map as it is built now won't fail 
with the addition sweep visitor.

I think the most likely problem is that I violated some kind of 
precondition requirement for the boolean operations code, resulting in 
an illegal shortcut in the boolean ops that produced a bogus map. :-( 
The map would pass is_valid but perhaps that test isn't conclusive?

cheers
Ben


Efraim Fogel wrote:
> Ben Supnik wrote:
> > Hi Y'all,
> >
> > I have an arrangement (built out of join and difference operations on 
> > polygons).  I am using an exact pred/exact constructions kernel and 
> > half -frounding-math on.  I am in debug mode, so I have all warnings, 
> > etc.
> >
> > After building my arrangement, I do an insert-curves on an 
> > axis-aligned bounding box that sometimes overlaps the horizontal and 
> > vertical edges of my map.
> >
> > If I insert the four sides using insert_curve, all is good.
> > If I insert all four using insert_curves, I get an assertion failure 
> > when trying to split a face.
> >
> > Now...looking at the code it seems clear that I should be using 
> > insert_curve anyway; doing a full copy of the arrangement to put only 
> > four curves in will be dominated by the copy cost.
> There are cases where applying incremental inserts is more efficient 
> than an applying an aggregate insert, especially when the number if 
> input curves is small, but which copy of arrangement do you refer to?
> > But...is there any case where the sweep-line-based insert_curves 
> > should blow up?  I am concerned that either my map is secretly invalid 
> > (it passes validation checks) or I have compiler settings such that my 
> > numerics are not stable.
> We are not aware of any bug of that sort. Try to come up with a test 
> case that exposes the problem, and that we can reproduce, and we'll look 
> at it.
> > cheers
> > ben

--
Scenery Home Page: http://scenery.x-plane.com/
Scenery blog: http://xplanescenery.blogspot.com/
Plugin SDK: http://www.xsquawkbox.net/xpsdk/
X-Plane Wiki: http://wiki.x-plane.com/
Scenery mailing list: 

Developer mailing list: