[Patch, Fortran] OOP: merging fortran-dev to trunk?

[Salvatore Filippone]

>Also, we should probably try to include additional checks, to make
>sure that stuff which will fail at runtime is already rejected at
>compile-time. However, I'm not quite sure how difficult this would be,
>or how much we would have to reject ...

I don't see how the situation of PR 41829 can be detected at compile
time, especially since the failing code will be most likely compiled
independently (and beforehand) of the code using it. 
I only have a vague idea of a runtime error  message, along these lines:
"The TBP cannot be found. This is a  limitation of current GFORTRAN
dynamic dispatching, see <whatever>" 

Salvatore