Re: defining an arbitrary type in your newer Fortrans

Aha, good to hear from the mouth of the horse.

Yep, once you open up the possibility of defining type aliases, you
have problems. Having worked on a Pascal compiler, I know there's a
lot of headaches in deciding what is type and assignment compatible
with what. Then since Fortran is big on arrays, you'd also want
paramaterizable types. And variable number of dimensions. And types
same as others, but with say the allocatable attribute. And so on.
Plenty of headaches, and lots of exceptions to document, and in a
language already carrying more baggage than Paris Hilton on a
round-the-world tour.

And since it's been possible to do write Fortran for 50 yrs now, one
can always say "we've never needed this before".

I don't quite see the problem with I/O-- if it could be done with the
base type, the aliased type should also I/O. At first blush.

.