IIDEX gives error in v 6.6B

rahzan · ‎09-20-2002

On using

routineX (line)
use numnerical_libraries, only: IIDEX
character(*), intent(in):: line
N=IIDEX(line, 'Hello')
...

I get the compiler error:

Error: If the actual argument is scalar, the corresponding dummy argument shall be scalar unless the actual argument is an element of an array that is not an assumed-shape or pointer array, or a subs

none of the inputs to IIDEX are arrays, or are they? I treied Line(:) but not good.

any ideas? else is there another case insensitive INDEX function?

Tim

Steven_L_Intel1 · ‎09-21-2002

Well, here's the interface for IIDEX from NUMERICAL_LIBRARIES.F90:

      integer function iidex (chrstr, key) 
      character, dimension(*), intent(in) :: chrstr 
      character, dimension(*), intent(in) :: key 
      end function

So it is indeed an array in the interface, but the interface is wrong, if I look up the documentation on IIDEX. It seems there are a few others like this, where some programmer at VNI thought that the above declaration was the same as character(len=*).

You could create your own interface for IIDEX and rename the one from NUMERICAL_LIBRARIES to be something else, so as to not conflict. That's what I would recommend.

Steve

rahzan · ‎07-03-2003

Yes, those airheads!

The same problem exists with IICSR, and likely other string manipulation routines.

An easy solution seems to be to feed the args to trim(), e.g.

n=IIDEX(trim(a),trim(b))

I can only guess how this might address the problem.

tim

Steven_L_Intel1 · ‎07-03-2003

I will make sure that this error is corrected for the 6.6C update (yes, still in the works.) We get the IMSL modules precompiled from VNI, but I went ahead and edited them to remove this class of error as well as a few others I observed.

I'll upload the corrected interfaces to our FTP site on Monday and provide a pointer here so that folks can try them.

Steve

Jugoslav_Dujic · ‎07-04-2003

It looks as if history repeats... similar error existed on several places in early DFWIN versions (probably inherited from Microsoft). For one, I recall that WIN32_FIND_DATA had file name declared as CHARACTER:: cFileName(MAX_PATH) -- it was a real PITA to do anything with it :-(.

Jugoslav