* It is recommended that a functions using the flags states so in its
* documentation.
*
- * There is another set of user data, namely a <tt>void *</tt>, for each
- * line, quad, etc. You can access these user pointers through the
- * functions <tt>user_pointer()</tt>, <tt>clear_user_pointer()</tt> and
- * <tt>set_user_pointer(p)</tt> in the accessor classes. These pointers are
- * neither used nor changed by the library and are @p NULL by
- * default. Thus, their handling is the sole responsibility of the
- * application program. Especially, the pointers are not inherited
- * to children upon refinement. Still, after a remeshing they are
- * available on all cells, where they were set on the previous mesh.
+ * There is another set of user data, namely a <tt>void *</tt>, for
+ * each line, quad, etc. You can access these user pointers through
+ * the functions <tt>user_pointer()</tt>,
+ * <tt>clear_user_pointer()</tt> and <tt>set_user_pointer(p)</tt> in
+ * the accessor classes. These pointers are not used nor changed in
+ * many places of the library, and those classes and functions that
+ * do use them should document this clearly; the most prominent user
+ * of these pointers is the Solutiontransfer class.
+ *
+ * The value of these user pointers is @p NULL by default. Note that
+ * the pointers are not inherited to children upon
+ * refinement. Still, after a remeshing they are available on all
+ * cells, where they were set on the previous mesh (unless, of
+ * course, overwritten by the SolutionTransfer class).
*
* The usual warning about the missing type safety of @p void pointers are
* obviously in place here; responsibility for correctness of types etc
* which does no harm since the vertices have a unique and continuous number
* unlike the structured objects lines and quads.
*
+ *
* @sect3{Boundary approximation}
*
* You can specify a boundary function for each boundary