J3/00-338
Date:      12 December, 2000
To:        J3
From:      Meadows, /interop
Subject:   Unresolved issues 218, 219, 238

Unresolved issue 218, page 400, asks whether a BIND(C) global variable
must be defined by C or whether it can be defined by Fortran.
It also asks for clarification of the linkage between the C variable
and the Fortran variable. Finally, it asks for rewording of shall
vs. may in Section 16.2.7 .

Unresolved issue 219, page 401, talks about needing details about
changes to the C variable affect the Fortran variable, and vice versa.

Unresolved issue 238, p. 67, asks if we should remove the prohibition
that BIND(C) variables cannot be initialized.

Subgroup believes that a global variable with the BIND(C) attribute
should cause that variable to exist if there
is no C definition of the variable. If there is no C definition
of the variable, and Fortran does not initialize the variable,
then the variable is initially undefined in Fortran.
If the variable is initially defined in Fortran, then there must
be no C definition of the variable. If the variable has a C
definition that initializes the variable, then it has the same
initial value in Fortran. The C and Fortran variables are
the same entity.

Edits:
[67:1-2] Delete "an object with the BIND attribute, an object in a named
COMMON block that has the BIND attribute,"

[67:3-8] Delete unresolved issue 238

[361:15] Add a new bullet item:
(5) Variables with the BIND attribute that are initialized by means other
than Fortran.
Add a note:
Note 14.x
module mod
	integer, bind(c,name="blivet") :: foo
end module mod

int blivet = 123;

In the above example, the Fortran variable foo is initially defined to
the value 123 by means other than Fortran.

[392:33] Add the sentences:
There does not have to be an interoperable C entity for a Fortran entity.
If there is such a C entity, it is the same entity as the Fortran
entity.

[400:35-44] Delete unresolved issue 218

[400:32] Change "shall be" to "is"

[400:34] Add the sentences:
A variable may be initially defined by Fortran
or by means other than Fortran but not both.

[401:35-40] Delete unresolved issue 219. Note: the last part of the
unresolved issue is resolved by Item 1 at [401:1-2].