gfortran, ncar and blockdata_test failures

[Walter Spector]

Jack Howarth wrote:
> I would think a decent argument could be made
> for it being a regression from g77.

It is certainly a regression.

The Standardness of the trick has been a long running subject of
debate.  No non-Standard syntax is involved.  It is the semantics
that are questionable.

In strict F66, BLOCK DATAs did not have names.  One often saw,
and linkers had to deal with, applications that had multiple unnamed
BLOCK DATA program units.  And if someone wanted to put them into a
library, how would you get them to link in?  So it was a very common
extension to allow naming of BLOCK DATAs.  Then, using the EXTERNAL
trick, make sure they got linked in from libraries.

F77 moved half way - allowing BLOCK DATA program units to be named.
This gave more credence to the trick - without specifically stating it.
But it left open-ended whether EXTERNAL could touch that name or not.

Nowadays, with modules, the fact that something gets USEd is a pretty
clear signal that the module needs to be linked in.

W.