Allocatable components to the right of a nonzero-rank part reference

[Brooks Moses]

FX Coudert wrote:
>>         (a) Allways reject (attached patch does that).
> 
> I really would like to say (a). We have enough bugs on valid code and  
> standard features missing, in addition to the slippery nature of  
> extensions.

Thirded.  I'm not fond of the idea of features that sometimes work and 
sometimes don't, and I agree that making this one work consistently in 
all cases is liable to be very complicated.

- Brooks