Why are there separate convectiondiffusion.hh and convectiondiffusionfem.hh?

The two seem to do the same, but convectiondiffusionfem.hh is the more complete implementation.

Should we deprecate the version in convectiondiffusion.hh?

"convectiondiffusion.hh" (used by "instationarytest.cc" and "nonlineardiffusion.cc") discretizes some non-linear CDE:

div( q(x,u) - D(x) v(u) grad w(u) ) = f(u)

whereas all other local operators for the CDE treat the linear case:

div( - D grad u + q u )

Maybe, we should rename it into "NonLinearCdeFEM".

Yeah, we should rename the file and the class inside this file.

Proposal for the filename: nonlinearconvectiondiffusionfem.hh.

@ango please switch to the convection with the variable intorderadd to be provided in the constructor of the local operator in convectiondiffusion.hh.

The correct integration order should be derived internally as in the file convectiondiffusionfem.hh and ```convectiondiffusiondg.hh````.

mentioned in merge request !30 (merged)

mentioned in commit f3350767

As both operators serve a very different purpose we should (at the moment) keep both and rename convectiondiffusion.hh should be renamed. On the long term perspective I think it could be reincorporated with an extended parameter class. But this is work we want to avoid right now.

mentioned in merge request !73 (merged)

Status changed to closed by commit 314b2643