[Patch, fortran] PRs 20862, 23152, 25053, 25063, 25064, 25066, 25067, 25068, 25069, 25307 data transfer constraints

[Salvatore Filippone]

I am surprised at Richard Maine's surprise about submodules.
I can see why submodules are high on the wish list, they sure are very
high on MY wish list (right after the allocatable extension). They are
the only sane mechanism around to separate interface from implementation
while still getting all the cross checks and private stuff access of a
"normal" module, and this is quite important for people like me who
often try out multiple implementations of the same interface, and would
rather avoid recompilation cascades. 

Just my € .02 
-- 
Salvatore Filippone -- sfilippone@NOSPAMuniroma2.it