cgal/Robustness/changes.txt

137 lines
3.9 KiB
Plaintext

1.23 (18 Sep 2003)
- more new headers
1.22 (18 Sep 2003)
- New headers
1.21 (10 Sep 2003) [rursu]
- Qt_help_window class is in the CGAL namespace now
1.20 (07 Sep 2003) [rursu]
- demo use now the new help class Qt_help_window instead of HelpWindow
- demo use the new icon demoicon
1.19 (14 Aug 2003) [rursu]
- fixed the demo for Windows platform
1.18 (12 Aug 2003) [rursu]
- modified Qt based demo. Added new features.
1.17 (09 Jul 2003) [rursu]
- replaced Leda based demo with Qt based demo
1.16 (17 May 2003) [af]
- Replaced Interval_base::unsafe_comparison with Interval_nt<false>::unsafe_comparison
1.15 (12 April 2003)
- Replace <strstream> by <sstream>.
1.14 (24 February 2003)
- Filtered_kernel only takes one template parameter now.
1.13 (14 September 2002)
- Use derivation instead of typedefs for the kernels to shorten the mangled
symbols (sylvain).
==================== CGAL 2.4 =========================
1.12 (4 December 2001)
- Remove #include to decrease package dependencies (lutz).
1.11 (27 September 2001)
- Changed CartesianLedaReal to ExactCartesian.
==================== CGAL 2.3 =========================
1.10 (1 August 2001)
- Use MP_Float when GMP is not there.
- Use Filtered_kernel instead of Filtered_exact for VC++.
- Remove warnings.
1.9 (31 July 2001)
- Last port to CGALWin.
1.8 (18 July 2001)
- Don't use <sstream> as it doesn't exist in GCC < 2.95.3.
1.7 (26 June 2001)
- Get rid of #include <LEDA/rat_point.h>, and use Cartesian_converter<>.
- Remove demo/Robustness/include/CGAL/kernel_to_kernel.h, as it's in CGAL.
- Port to CGAL window.
- Don't use obsolete Kernel::Less_distance_to_point_2.
1.6 (20 June 2001)
- Update to new header style, and simplify code.
- Try to use Quotient<MP_Float> in place of leda_real.
1.5 (20 Sep 2000)
- workaround for VC++ added in triangulation_of_intersection_points_2:
#ifdef CGAL_CFG_MATCHING_BUG_2
#define CGAL_IA_CT double
#define CGAL_IA_PROTECTED true
#define CGAL_IA_ET leda_real
#define CGAL_IA_CACHE No_Filter_Cache
#endif
#include <CGAL/Arithmetic_filter.h>
as suggested by Sylvain Pion
1.4 (01 Aug 2000)
- C-array with non-constant size in orientation.C moved to dynamic
heap, i.e. allocated using new
1.3 (17 Jul 2000)
- std::atoi replaced by CGAL_CLIB_STD::atoi
- missing std:: added to some back_inserter calls
- adjustments to changes in interval arithmetic
o Interval_nt<true> used instead of Interval_nt
o Interval_base::unsafe_comparison used
1.2 (18 May 2000)
- README file added
- orientation_2.C and orientation_IA_2.C added to demo/Robustness
they count how often orientation test computation is successful
with double and with interval arithmetic, respectively.
1.1.2 (08 May 2000)
- std:: added to cin
1.1.1 (19 Apr 2000)
- std:: added to back_inserter
- CGAL_CLIB_STD:: added to atoi
- decls moved out of for scope to please VC++
- CGAL_USE_LEDA checked
1.1 (18 Apr 2000)
- hull_of_intersection_points_2.C in demo/Robustness
computes the convex hull of intersection points of random
line segments using doubles and leda_reals. Shows points
incorrectly classified as extreme points by the double-based
computation
- triangulation_of_intersection_points_2.C in demo/Robustness
computes the Delaunay triangulation of intersection points of
random line segments using leda_reals, Filtered_exact< double,
leda_real> and double. With doubles you often get assertion
violations. By altering the failure behaviour of CGAL, the
assertion violation is shown in the window. (Also, sometimes
the Delaunay code run into an infinite loop.)
1.0.1 (21 Mar 2000)
- packaging script fixed
- file header fixed
1.0 (21 Mar 2000)
- intersection_points_on_segment_2.C in demo/Robustness
creates segments with integral endpoints and checks whether
intersection point lies on segments using different kernels