sign_of_determinant2x2
sign_of_determinant3x3
sign_of_determinant4x4
sign_of_determinant5x5
sign_of_determinant6x6
to:
sign_of_determinant
So that we have less dimension-dependent namings, at least internally...
It makes more code parameterized by the dimension as template argument.
It provides iterators and a few basic functions already such as operator==...
I added CGALi::make_array() to easily and efficiently construct them.
* CGAL_error to CGAL_error_msg
* introduced a macro CGAL_error()
* added some words about CGAL_error to the developers manual
* renamed most of assert(x) into CGAL_assertion(x)
* renamed exit(x) with x != 0 , CGAL_assertion(false) and assert(false) into CGAL_error
* CORE left untouched, OpenNL changed
of "forwarding constructors".
Quoting some comment in the code:
"
This is a simple tag which is used as additional (first) argument in
some kernel functors, to tell them to return the base (rep) class,
instead of the main type (e.g. Kernel_base::Point_2 instead
of Kernel::Point_2). This is a minor optimization which prevents
useless copies of the "reps".
Those functors are only those used in the constructors of the kernel
types like Point_2, so it's limited.
The real solution will be to use "forwarding constructors", when they
will be available in C++.
In the mean time, this should be a mostly/hopefully internal hack.
"
Construct_point_2 functors be the Rep class (the base).
This avoids conversions Rep -> Point_2 -> Rep, hence
useless copies of objects.
The result_type of the functors does not change
(we therefore return a type which is only convertible
to result_type, but hopefully this is fine, and what standard
requirements on functors are anyway).
A real fix for this would require the language addition of
"forwarding constructors".
used by Cartesian_converter/Homogeneous_converter, where we
absolutely must not throw.
TODO : find a way to keep the assertion while having the
converters still work. (should types come with their
own conversions, like they all have .bbox() ?)
The additional dummy "int" specifies that the 2 points are
the lower-left and upper-right corner. This is more efficient
when one knows they are already in this configuration.
Same thing for Iso_cuboid_3, and the functors.
Use them in Cartesian_converter and Homogeneous_converter.
Sylvain Pion
Andreas Meyer
Andreas Fabri