CGAL users discussion list

[Andreas Fabri <>]

pong!!

Hi Matt,

We will look into it, your analysis might be correct, but we are
currently busy with higher priority tasks.

andreas



On 27/03/2012 18:34, Matt Johnson wrote:
> Ping?
>
> On 03/21/2012 03:31 PM, mrj10 wrote:
> > I was looking at Interval_skip_list.h and noticed what I think might
> > be a bug
> > in adjustMarkersOnDelete(). Many places in the code need to do
> > special-case
> > check to see if the source or sink of an ISL edge is the ISL header or
> > footer, but this check is not done in the if statement on line 860 (in
> > CGAL
> > 4.0). The check is currently:
> >
> > if(x->forward[i]==0 ||
> > !m->getInterval()->contains_interval(update[i]->key, x->forward[i]->key))
> > {
> > newDemoted.insert(m->getInterval());
> > }
> >
> > If update[i] is the header, update[i]->key will be garbage (we don't
> > initialize the header's key in the ISL constructor), so the
> > contains_interval() call may do something unexpected. It seems like the
> > check should be:
> >
> > if(x->forward[i]==0 ||
> > isHeader(update[i]) ||
> > !m->getInterval()->contains_interval(update[i]->key, x->forward[i]->key))
> > {
> > newDemoted.insert(m->getInterval());
> > }
> >
> > The implementation is very clever in other places, so it's entirely
> > possible
> > I may have missed an invariant that makes this check unnecessary, but
> > I've
> > run into bugs where edges from the header have markers on them, and they
> > appear to be caused by a bad result from this check.
> >
> > Best,
> > Matt
> >
> > --
> > View this message in context:
> > http://cgal-discuss.949826.n4.nabble.com/Interval-Skip-List-bug-tp4493660p4493660.html
> >
> > Sent from the cgal-discuss mailing list archive at Nabble.com.


--
Andreas Fabri, PhD
Chief Officer, GeometryFactory
Editor, The CGAL Project

phone: +33.492.954.912    skype: andreas.fabri