a whole section 2.x for it.}
\item changes of interface of AffineConstraints
\todo[inline]{Wolfgang to write}
-\item We have clean up the orientation definition of faces in (2D and) 3D. Three
- Booleans used to be used to define it for hexedras: \textit{orientation},
- \textit{flip}, and \textit{rotation}. While introducing simplex and mixed
- mesh support in \dealii, we merged these three into a single variable of type
- char internally. Now, we also expose these to the outside, leading to
- multiple incompatibilities, since the template type of different data
- structures has changed and the number of input/output arguments have decreased.
- \todo[inline]{DavidW: Please describe some details here.}
-\item We have performed various performance improvements throughout the library,
- which we detail in the following sections.
+ \item \dealii{} internally needs to keep track of the relative
+ orientations of the coordinate systems associated with neighboring
+ cells, as well as of the coordinate systems of faces and edges
+ within the coordinate system of the cells they are part
+ of. Historically, different places within the library that had to
+ do that had grown their own, independent ways of representating
+ this information, much of which was difficult to understand.
+ We have now cleaned up this area. Specifically, instead of using three
+ Booleans to define relative orientations for hexedra (\textit{orientation},
+ \textit{flip}, and \textit{rotation}), we have internally merged these three values into a single variable of type
+ \texttt{char} -- i.e., the smallest integer data type that has at
+ least three bits to represent the information. Because the way to represent
+ orientations is part of the public interface, the change causes
+ some incompatibilities in types and function signatures.
+ \todo[inline]{DavidW: Please describe some details here. -- When
+ done, please remove this comment.}
\item The \texttt{SolverGMRES} class now offers a third orthogonalization
method, the classical Gram--Schmidt method with delayed orthogonalization
\cite{Bielich2022}. Furthermore, the solver now specifies the maximal basis