bug with internal file I/O?

[Magnus Hagdorn]

On Fri, 2007-06-08 at 08:00 -0700, Steve Kargl wrote:
> 
> Although the program is invalid, there is definitely some problem
> with the implied-do loop in the IO statement.  Consider this valid
> Fortran program
> 
> program test
>    implicit none
>    character(len=100) :: value
>    integer, dimension(100) :: intvalues
>    integer i
>  
>    i = 2
>    intvalues = 42
>    value = "2 5 69"
>    write(*,*) len(trim(value))
>    read(value,*,end=20,err=20) (intvalues(i),i=1,100)
> 20 i=i-1
>    !
>    ! This should print 1 2 5 69
>    !
>    write(*,*) i, intvalues(1:3)
>    !
>    ! This prints the expected output 1 1 100
>    !
>    i = 1
>    intvalues = (/ (i,i=1,100) /)
>    print *, i, intvalues(1), intvalues(100)
> end program test
> 
> The problem is a scoping issue.  The implied-do-index has the 
> scope of the implied-do.
> 

ok, I can see the problem. Unfortunately, it doesn't solve what I had
hoped for. I wanted a way to find the number of integers I just read
from the string. I guess I could initialise the integer array to some
unlikely value and check for its first occurrence.

> Magnus, can you file a bug report in bugzilla?
> 

I just created a bugzilla account and will report the bug ASAP.

Cheers
magnus