[dune-functions] Interface limitations in the functions interface

[Carsten Gräser]

Am 19.03.2015 um 21:13 schrieb Christian Engwer:
> OK, so some further insights...
> 
>>> A)I have the impression that it should be not to complicated
>>>  to implement this by extending the current interface with
>>>  varidadic templates. But I'll have to look into the details.
>>>  On thing that I'd really like to avoid is supporting a
>>>  total derivative wrt all domains at one then, because it
>>>  will in general be hard to define a proper range type for this.
>>
>> I started a prototype implementation for this... it is feasible, although not totally straight forward
> 
> it is owrking now, but only for the derivatives, I didn't yet look
> into anything like localView, or bind.
> 
> I seems to be not too complicate :-) if you want, you can have a look
> at the feature/multi-argument-functions branch.
Looks nice. Unfortunately it seems that you did also not find
a way around the hierarchies to add all the derivative methods.

> Besides these experiments, I think we can work around the problems in
> dune-pdelab, at least if we don't use the virtual interface. For the
> virtual interface, it would be good to have a hook in the interface
> wrapper, which allows to inject other interfaces. I have an idea about
> how to do this, but I didn't test it yet. This owuld allow us to reuse
> major parts of the code, derive our own typeerasure class from
> DifferentiableFunction and be able to inject the setTime method into
> this wrapper.
I also wanted to do this for GridFunction and LocalFunction, but
did not find a working solution. But I can't remember the reason
now. Making the base class a template parameter of the wrapper
should work.

Best,
Carsten