CGAL users discussion list

[Marc Glisse <>]

On Wed, 25 May 2011, GIS Worx wrote:

> Increasing the stack size to 20MB from 1MB does not resolve the issue.

Did you check that the way you pass the option works? (usual way to do 
that is to try with an example that works and reduce the stack size to 
something ridiculously small to force an overflow) I am asking because the 
internet is full of reports that /F only works in very specific 
circumstances whereas /STACK works more often, or the reverse. I don't use 
that compiler so I wouldn't know.

20MB sounds bad, it indeed seems to indicate an infinite loop. Do you have 
a longer stack trace in the 20MB case?

> The lazy kernel seems to have a DAG underlying it. What checks enforce 
> that it is a DAG, i.e. is it possible I've got a long loop in my 'DAG'?

Well it seems hard from the code to produce a loop (but you never know).

> Is there a construction I can use besides Lazy_exact_nt which will side-step
> this issue? I want to use a Exact_predicates_exact_constructions_kernel, but
> trying all combinations of using the flag CGAL_DONT_USE_LAZY_KERNEL with
> CGAL_USE_GMP turned on and off, I run into ambiguous call errors with
> CGAL::internal::fit_in_double in Orientation_2.h.

Maybe related to this?
https://lists-sop.inria.fr/sympa/arc/cgal-discuss/2011-01/msg00049.html

If you get different errors, it's worth reporting them.

--
Marc Glisse