Re: [boost] Formal Review: Boost.Polygon starts today August 24, 2009

Subject: Re: [boost] Formal Review: Boost.Polygon starts today August 24, 2009
From: Andreas Fabri (andreas.fabri_at_[hidden])
Date: 2009-08-27 15:43:58

Simonson, Lucanus J wrote:
> Andreas Fabri wrote:
>> Hi Luke,
>>
>> thank you for your explanations for the 45 degree case. You have a
>> nice solution. I guess for the general case I have to read the
>> BoostCon paper first. 
>>
>> andreas
> 
> Thank you for saying so.
> 
> As requested, I haved added a test case for which CGAL fails to subversion
> 
> https://svn.boost.org/svn/boost/sandbox/gtl/forcgal.cpp

Hi Luke,

Using CGAL with just floating point arithmetic is rather nonsense,
especially when you test robustness.

I guess I need a recent version of boost, plus sandbox/gtl,
in order that it compiles?

Best regards,

andreas

> 
> I appologize that the test case is rather large.  Expected output is:
> 
>  $$$$$$$$$$$$ RANDOM TEST $$$$$$$$$$$$$$$$
> Square BOXX size: 6000 *************************
> CGAL error: precondition violation!
> Expr: compare_xy(cv1.right(), p) == LARGER && compare_xy(cv2.right(), p) == LARGER
> File: /nfs/ltdn/disks/tcad_lmg_db_exchange_01/CGAL-3.3.1/include/CGAL/Arr_segment_traits_2.h
> Line: 609
> Explanation:
> terminate called after throwing an instance of 'CGAL::Precondition_exception'
>   what():  CGAL ERROR: precondition violation!
> Expr: compare_xy(cv1.right(), p) == LARGER && compare_xy(cv2.right(), p) == LARGER
> File: /nfs/ltdn/disks/tcad_lmg_db_exchange_01/CGAL-3.3.1/include/CGAL/Arr_segment_traits_2.h
> Line: 609
> Abort
> 
> This assertion in CGAL code looked to me like some condition that should be ensured by CGAL itself and is an internal error rather than a precondition violation.  My interpretation was that the wrong exception type was used.  I could be wrong, however, and if you find there was in fact something wrong with the polygon I produced I will eagerly fix my algorithm.  My intention was to implement a booleans algorithm with the weakest possible preconditions and the strongest possible postconditions.
> 
> Thanks,
> Luke
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
> 
>