songsenand
/
cgal
mirror of https://github.com/CGAL/cgal
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Reworded the warning about ExactAlphaComparisonTag

This commit is contained in:
Mael Rouxel-Labbé 2017-06-13 14:11:37 +02:00
parent 6b524ac88a
commit bcdfaefee9
2 changed files with 22 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
Alpha_shapes_2/doc/Alpha_shapes_2/CGAL/Alpha_shape_2.h
View File

@ -28,18 +28,21 @@ Note that `Dt::Geom_traits`, `Dt::Vertex`, and `Dt::Face` must be model the conc
\link Tag_true `Tag_true`\endlink, triggers exact comparisons between alpha values. This is useful
when the underlying triangulation is instantiated with an exact predicates inexact constructions
kernel. By default the `ExactAlphaComparisonTag` is set to \link Tag_false `Tag_false`\endlink as it induces a small
overhead. Note that since such a strategy does not make sense if used together with a traits class with exact constructions,
the tag `ExactAlphaComparisonTag` is not taken into account if `Dt::Geom_traits::FT` is not a floating point number type.
overhead.
\warning The tag `ExactAlphaComparisonTag` is currently ignored (that is, the code will
\warning The tag `ExactAlphaComparisonTag` is currently ignored (meaning that the code will
behave as if `ExactAlphaComparisonTag` were set to \link Tag_false `Tag_false`\endlink)
if the traits defines a point type that is not implicitely convertible to the two-dimensional point type
of a CGAL kernel. This is because we internally use the class `Cartesian_converter` to switch
between the traits and exact CGAL kernels. <br>
This means for example that it is useless to use the tag `ExactAlphaComparisonTag`
in conjonction with the traits class `CGAL::Projection_traits_xy_3`: the latter
camouflages a `CGAL::Point_3` as a `%Point_2` but the exact kernel `EK`
will not know how to convert from the camouflaged `CGAL::Point_3` to `EK::Point_2`.
if either condition is met:
- `Dt::Geom_traits::FT` is not a floating point number type. Indeed, this strategy
does not make sense if the traits class already provides exact constructions.
- the traits defines a point type that is not implicitely convertible to the two-dimensional point type
of a CGAL kernel. This is because we internally use the class `Cartesian_converter` to switch
between the traits class and exact CGAL kernels. For example, `ExactAlphaComparisonTag`
will be ignored if the traits class `CGAL::Projection_traits_xy_3` is used since
this traits class camouflages a `CGAL::Point_3` as a `%Point_2` and `Cartesian_converter`
does not know how to convert from the camouflaged `CGAL::Point_3` to the two-dimensional
point of an exact kernel.
\cgalHeading{I/O}

15
Alpha_shapes_3/doc/Alpha_shapes_3/CGAL/Alpha_shape_3.h
View File

@ -23,14 +23,17 @@ must be model the concepts `AlphaShapeTraits_3`,
\link Tag_true `Tag_true`\endlink, triggers exact comparisons between alpha values. This is useful
when the Delaunay triangulation is instantiated with an exact predicates inexact constructions
kernel. By default the `ExactAlphaComparisonTag` is set to \link Tag_false `Tag_false`\endlink as it induces a small
overhead. Note that since such a strategy does not make sense if used together with a traits class with exact constructions,
the tag `ExactAlphaComparisonTag` is not taken into account if `Dt::Geom_traits::FT` is not a floating point number type.
overhead.
\warning The tag `ExactAlphaComparisonTag` is currently ignored (that is, the code will
\warning The tag `ExactAlphaComparisonTag` is currently ignored (meaning that the code will
behave as if `ExactAlphaComparisonTag` were set to \link Tag_false `Tag_false`\endlink)
if the traits defines a point type that is not implicitely convertible to the three-dimensional point type
of a CGAL kernel. This is because we internally use the class `Cartesian_converter` to switch
between the traits and exact CGAL kernels.
if either condition is met:
- `Dt::Geom_traits::FT` is not a floating point number type. Indeed, this strategy
does not make sense if the traits class already provides exact constructions.
- the traits defines a point type that is not implicitely convertible to the three-dimensional point type
of a CGAL kernel. This is because we internally use the class `Cartesian_converter` to switch
between the traits class and exact CGAL kernels.
Note that this class is used for <I>basic</I>, <I>weighted</I>,
and <I>periodic</I>Alpha Shapes.