forums.silverfrost.com

[TarHot]

Dear forists:

A Block Data routine is supossed to initialize the common areas in
DELSML.INC. Both are included below.

FTN95 complains with the following disgnostic:

1025) DATA MSKTPA / Z80000000 /
*** All items in a a DATA statement's value list must be literal constants

Could you please let me know why DATA item values cannot be integers, or
hexadecimal, as defined in my INCLUDE file?

Thanks a lot


---------------------- DELSML.INC --------------------------
C Include file with parameters and arrays for the smoothing and
C sharpening factors. Also bits indicating active status and fault
C status for a triangle.
C Revision List:
C March 1994 Original Version
C
C ... Sharpness codes for the 'fast' smoothing operation. Actual
C values must be loaded by including BLOCK DATA program LOADSH.
C
INTEGER * 4 SHLE
PARAMETER (SHLE = 4)
C
CHARACTER * 1 SH(SHLE)
C
INTEGER * 4 MSKTPA ,MSKTFL
INTEGER * 4 MSKTP1 ,MSKTP2 ,MSKTP3
INTEGER * 4 MSKTPS(3)
EQUIVALENCE (MSKTP1 ,MSKTPS(1))
EQUIVALENCE (MSKTP2 ,MSKTPS(2))
EQUIVALENCE (MSKTP3 ,MSKTPS(3))
C
INTEGER * 4 ISHTR(3)
COMMON /SHINT4/MSKTPA ,MSKTFL ,MSKTPS ,ISHTR
COMMON /SHCHAR/SH
CCC

--------------------------------BLOCK DATA-----------------------------

C DELSHL BlockData Initialization of status masks and common areas
BLOCK DATA
C For initializing status masks and smoothing and sharpening common area
C in file DELSML.INC
C Originally: INCLUDE '/u/albert/dvlp/newup/DELSML.INC'
INCLUDE 'C:\users\albert\AIX\dvlp\newup\DELSML.INC'
C
DATA SH(1) / 'v'/
DATA SH(2) / 's'/
DATA SH(3) / 'h'/
DATA SH(4) / 'b'/
C
DATA MSKTPA / Z80000000 /
DATA MSKTFL / Z01000000 /
DATA MSKTPS(1)/ Z60000000 /
DATA MSKTPS(2)/ Z18000000 /
DATA MSKTPS(3)/ Z06000000 /
C
DATA ISHTR(1)/ 29 /
DATA ISHTR(2)/ 27 /
DATA ISHTR(3)/ 25 /
C

[LitusSaxonicum]

Possibly because your hexadecimal constants are not in the FTN95 form?

(Page 183 of the FTN77 manual - see the documentation section of this website - says they should be Z'80000000' ).

Seems probable to me, as Z80000000 is a valid, if rather odd, symbolic name.

Never used this type of specification, so can't be sure.

E

[TarHot]

Many thanks. You hit the nail in the right place.

[TarHot]

Many thanks. You hit the nail in the right place.

[klclark]

WARNING - Common block "//" was previously defined as size 8704 but is now defined as size 8656
WARNING - Common block "//" was previously defined as size 8704 but is now defined as size 24

How do I fix this or does it matter. I have data overlaying data after calling a function subprogram.

[LitusSaxonicum]

You can ignore warnings, and you can also set the compiler to not issue specific warnings (using -ignore followed by the warning number). You could pad out this block in each routine where the warning crops up with extra bites to make it the same length, but only if the warning bothers you.

Eddie

[Wilfried Linder]

I prefer not to ignore valuable warnings like these. A possible reason:

[IanLambley]

Create an "INCLUDE" file with all the variable/array definitions and common block definitions. Then include it in the appropriate routines.
Or do it via modules and "USE"
Ian

[LitusSaxonicum]

Wilfried,

He already said he was overlaying data - i.e. he knows that the COMMON block contains different things in different places.

I always exclude Warning 179 because it irritates me (it is the one that tells me I compared something to zero) - because I always test for zero before dividing by something that could be exactly zero as a result of a data error. Agree you are unlikely to get exactly zero as a result of a calculation, but then I wouldn't test for zero, I'd test for a small range around zero.

Ian,

If he uses INCLUDE, he has to INCLUDE different things in the different subroutines because of how he says that he has overlaid different things into this particular COMMON.


TarHot,

Why do you need to overlay anything? provided that you aren't a user craving for multi-gigabyte arrays, then there is ample space for just about anything - especially if you are using it for small (c. 8k bytes) COMMON.

Eddie

[klclark]

I have data overlaying data and I do not want that. Sorry for the misunderstanding. Thanks for the help, I'm sure I'll have more questions.

[klclark]

I'm trying to use common in a function as well as a subroutine. I looks like everything is OK until the call to the function or the subroutine. I get a run time error when trying to print either argument "S" or "T". I tryed looking at your BLOCK COMMON but I couldn't find enough information to understand how to set it up. I think I just need a very good example. Can you help? Is there something I need to do in the SLINK? How would I set that up?


[code]
winapp
program testprint02
INCLUDE <windows.ins>
EXTERNAL printer

INTEGER S(7),T(7)
CHARACTER *4 LINEA

COMMON S,T,LINEA

DO I=1,7
S(I) = I
T(I) = I
ENDDO

LINEA = 'TEST'
!
DO I = 1,7
PRINT *,S(I),T(I)
ENDDO
PRINT *,s(1),t(7),LINEA

i=winio@('%sc[Ready to Print?]%ww[invisible,toolwindow]','PRINTER_OPEN',7,printer)

call testit
close (7)

END
! CONTINUE
! CONTAINS

! c-----
INTEGER FUNCTION printer()

COMMON LINEA,S,T

write(7,15)'Hello'
15 format(1x,a5)
do i = 1 , 10
print 10, i
10 format (5X,i5)
write (7,10) i
end do
PRINT *,s(2),LINEA,'FUNCTION'
printer=0
END FUNCTION PRINTER

subroutine testit

COMMON LINEA,S,T

print *,'subroutine testit'
write (7,20)
20 format (1x,'subroutine testit')
PRINT *,LINEA,'SUBROUTINE'


end SUBROUTINE TESTIT

! END PROGRAM TESTPRINT02

[brucebowler]

It's been a long time since I've used common, but the entries are order dependent, not name dependent thus...

[klclark]

Well I changed the order and I still get a run time error.

"Error 29. Call to missing routine _S at ox004014e3

I also tried to use a CONTAIN statment and that screws up the Function Subroutine.

HELP!

[LitusSaxonicum]

LINEA needs to be declared as CHARACTER*(4) in every routine where it is used: the definition in the calling program is not enough. S and T are arrays, and their lengths also need to be declared in every routine.

This is because a variable name has a scope that is limited to the routine in which it is declared and used. Up to Fortran 77 there was no way of giving a variable name that "spanned across" routines. Under Fortran 90 and later, a variable name can have a wider scope using CONTAINS and/or MODULEs. In my opinion, you have to use this approach OR the approach using COMMON, because if you mix them you will get in a terrible mess. COMMON is, in any case, a tricky thing to use.

Some people will also advise you to use the declaration:

[klclark]

Please bare with me on this, I've been working with Fortran since 1967 and used F77 thru 1988. The F77 was written by Dartmounth College capable of accessing HISAM data bases. This F77 used common the way it should be used which really ment common in all programs, included programs and subroutines. I agree that not all Fortrans are created equal but since I've been out of contact for about 30 years I have a little learning curve to go thru. The basics are still the same, it's just the new stuff that's been added over the years that I missed out on. The biggest change is the Windows interface. Please hang in there with me.

I may just abandon the Common stuff and use GOTO.