#597 Make RefFEltSpace a class of the Private namespace. Make also the constructor of ElementaryData<> private: only GlobalVariationalOperator<> base class should be allowed to instantiate one.

#185 Rename UnknownAndNumberingSubset ExtendedUnknown (as shape function label will also be defined there).

#185 Rename UnknownAndNumberingSubset ExtendedUnknown (as shape function label will also be defined there).

#557 Report the changes of compilation in SCons. SCons doesn't compile until the end: it is assumed at the moment that all the model instances are broken (it is up to #556 to fix it; there won't by any tag until this ticket is performed).

#534 Rename MatrixVectorNature OperatorNS::Nature, which is more appropriate (with the exception of boundary conditions, for which it is a bit awkward). Move ElementaryData to LocalVariationalOperators group, which is much more sensible.

#506 Move a folder within operators: Private namespace got content related only to GlobalVariationalOperator so its place was with brand new dedicated folder.

#506 GlobalVariationalOperator: introduce policy to replace the newly introduced template parameters NunknownT and IsMonolithicT. Policies will be used to store helper local matrix or vector.

#496 Introduce an object that stores the matrix pattern and two numbering subsets (one for rows and one for columns). It does not have yet its final form: it is a step toward the full introduction of non square matrices.

#499 The number of processor-wise dofs per numbering subset is now computed and stored; this is what is displayed in the output files.

#480 Write dof output taking into account the numbering subset. PostProcessing however has been quickly patched by hardcoding the fact there is only one numbering subset which index is 1; ticket #495 has been issued to fix properly PostProcessing.

#480 Remove an obsolete class, and move the numbering subset related ones from FiniteEltSpace to Unknown folder.

#480 Keep slitting older ComputeMatrixPattern operations: now computation of the number of dofs is performed prior to ComputeMatrixPattern() call.

#480 Split the creation of the matrix pattern to the actual reduction to processor-wise. Not yet done: SequentialCase and ParallelCase should be completely merged.