|
|
 |
 |
 |
 |
Fortran Programming Language
|
 |
 |
 |
 |
|
|
|
|
A question about "intent(out)"?
Bellow is a subroutine in my program. I use
"integer(4),intent(out)::g(:)" but did not give it shape(Note: g(:) is
not a global variable). However, the program can give right redults,
why?????
subroutine num_to_g(num, nf, g)
implicit none
integer(4),intent(in)::num(:), nf(:,:)
integer(4),intent(out)::g(:)
integer(4)::i, k, nod, nodof
nod = ubound(num,1)
nodof = ubound(nf,1)
do i = 1, nod
k = i*nodof
g(k-nodof+1:k) = nf(:,num(i))
end do
end subroutine num_to_g
On Jun 1, 5:41 am, li.sim@gmail.com wrote:
> Bellow is a subroutine in my program. I use > "integer(4),intent(out)::g(:)" but did not give it shape(Note: g(:) is > not a global variable). However, the program can give right redults, > why????? > subroutine num_to_g(num, nf, g)
> implicit none
> integer(4),intent(in)::num(:), nf(:,:)
> integer(4),intent(out)::g(:)
> integer(4)::i, k, nod, nodof
> nod = ubound(num,1)
> nodof = ubound(nf,1)
> do i = 1, nod
> k = i*nodof
> g(k-nodof+1:k) = nf(:,num(i))
> end do
> end subroutine num_to_g
x(k:,:) etc.), then the shape of the dummy argument is inquired from
the associated actual argument, i.e. the assumed shape arrays carry
their dimensions with them.
The problem is that I didn't use "intent(in)" but "intent(out)". This
means g(:) is no relation with the any input values. And g(:) can be
only assgin a value in the subroutine itself, however,g(:) didn't has
a shape from the first. So, if I assgin g(1)..g(10),then the output
g(:) is a array contain 10 element; or if I assgin g(1)..g(100),then
the output g(:) is a array contain 100 element.This,I'm puzzeled???
li.sim @gmail.com wrote: > The problem is that I didn't use "intent(in)" but "intent(out)". This > means g(:) is no relation with the any input values. And g(:) can be > only assgin a value in the subroutine itself, however,g(:) didn't has > a shape from the first. So, if I assgin g(1)..g(10),then the output > g(:) is a array contain 10 element; or if I assgin g(1)..g(100),then > the output g(:) is a array contain 100 element.This,I'm puzzeled???
calling routine.
If you use an ALLOCATABLE dummy argument then you can
determine the shape in the called routine and ALLOCATE the
array as needed.
-- glen
| The problem is that I didn't use "intent(in)" but "intent(out)". This
| means g(:) is no relation with the any input values. And g(:) can be
| only assgin a value in the subroutine itself, however,g(:) didn't has
| a shape from the first. So, if I assgin g(1)..g(10),then the output
| g(:) is a array contain 10 element; or if I assgin g(1)..g(100),then
| the output g(:) is a array contain 100 element.This,I'm puzzeled???
No, you get it wrong. For an assumed-shape array dummy, intent(OUT)
specifies that the *contents* (values) of the array are to be filled
by the routine, and not used within it before it defines them. However,
array's *shape* (rank and dimensions) are still inherited from the
actual argument.
Note that INTENT never/(actually, rarely*) has any effect on the semantics
of produced code; in other words, the generated machine code will be the
same with or without it. It only imposes restrictions on what you
may do within the code and improves readability.
*) the exceptions I recall are allocatable array dummies and
user-defined types with default initialization. In those cases,
the argument is "cleared" (deallocated/default-initialized) on the
entry to the routine if INTENT(OUT) is specified.
--
Jugoslav
___________
www.xeffort.com
Please reply to the newsgroup.
You can find my real e-mail on my home page above.
On May 31, 9:41 pm, li.sim@gmail.com wrote:
> subroutine num_to_g(num, nf, g) > implicit none > integer(4),intent(in)::num(:), nf(:,:)
> integer(4),intent(out)::g(:)
> integer(4)::i, k, nod, nodof
> nod = ubound(num,1)
> nodof = ubound(nf,1)
> do i = 1, nod
> k = i*nodof
> g(k-nodof+1:k) = nf(:,num(i))
> end do
> end subroutine num_to_g
function as your subroutine.
|
|
|
|
|
|
Add to del.icio.us |
Digg this |
Stumble it |
Powered by Megasolutions Inc