It seems that the standard (or at least the standard literature) is not clear
on this one, but Sun and Intel compilers agree that it's good.
The problem we have is actually about a derived type that gets two different
alias names in USE/ONLY statements. In the end, it should be treated as two
different types but it is really one implementation.
It works the same with plain integers, so I use these for illustration.
Example program:
module base
integer :: baseint
end module
module a
use base, ONLY: aint => baseint
end module
module b
use base, ONLY: bint => baseint
end module
module c
use a
use b
private
public :: aint, bint
end module
program user
use c, ONLY: aint, bint
aint = 3
bint = 8
write(*,*) aint
end program
gfortran (4.1, 4.2, 4.3), and incidentally also g95 (if that interests
anyone;-) refuse this with:
public :: aint, bint
1
Error: ACCESS specification at (1) was already specified
If you just remove the private/public stuff in module c, like that:
module c
use a
use b
end module
The code compiles fine and I get the expected output fromt the main program:
8
...which is the value of baseint, set through bint and read through aint.
It feels like gfortran is too eager in error checking there on harmless code. I
am not sure if this code is strictly Fortran90 conforming... but I read that a
variable is allowed to have different aliases; just didn't find mentioning of
one module making public several of these.
--
Summary: too cautionous alias checking with renamed
variables/types from modules (Error: ACCESS
specification already specified)
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: thomas dot orgis at awi dot de
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36454
