VARARGS(3F)VARARGS(3F)NAME
varargs, argmnt, getadr, nullok, xetarg, retour - allow variable number
of arguments in argument list
SYNOPSIS
subroutine argmnt( nargs )
integer*4 nargs
subroutine getadr( n, iaddr )
integer*4 n, iaddr
integer*4 function nullok( n, iaddr )
integer*4 n, iaddr
subroutine xetarg( n, len, iarg )
integer*4 n, len
anytype iarg
subroutine retour( nargs, len1, val1, len2, val2, ..., lenn, valn )
integer*4 nargs, len1, len2, ..., lenn
anytype val1, val2, ..., valn
DESCRIPTION
These utilities are used to provide f77 support for subroutines with
variable number of arguments. In order to use these utilities, all
variable argument subroutines must be declared in each source file before
they are referenced or defined. This is done by adding a $varargs
compiler directive at the beginning of the source file.
For example:
$varargs vasub1 vasub2 vasub3
where vasub1, vasub2, and vasub3 are the names of the variable argument
subroutines which are referenced or defined in the current source file.
argmnt returns the number of actual arguments in the integer variable
nargs. The default behavior is to count each character argument in the
actual argument list as two arguments since both the character address
and its length will be put on the argument stack. The -chararg1 option
can be used to count each character argument as only one argument.
getadr returns the address of the nth argument in the variable iaddr.
This function has to be used to get the length of a character argument
since it is passed by value, not by reference like other Fortran
arguments.
nullok returns 0 if the address of the nth argument is the same as the
address contained in the variable iaddr.
Page 1
VARARGS(3F)VARARGS(3F)
xetarg initializes len bytes of iarg with the value of the nth argument .
retour returns values to the calling program by setting nargs actual
arguments using len1, len2,..., lenn bytes of the values stored in the
variables val1, val2,..., valn, respectively.
In normal usage of variable argument subroutines, when the formal
argument list is specified using the maximum number of arguments the
subroutine can received, the argument addresses and return values are
passed in the standard f77 convention. The only utility needed in this
case is argmnt to determine the number of actual arguments passed to the
variable argument subroutine.
The other utilities are needed when there is no formal argument list in
the variable argument subroutine. xetarg is normally used to initializes
some local variables to the initial values of the actual arguments.
retour is then used to return the locally calculated values back to the
calling subroutine by setting the actual arguments to the values of the
local variables.
Example:
$varargs chsign
program tvararg
i = 1
j = -2
k = 3
call chsign( i, j, k, "fourth argument", "fifth" )
print *, i, j, k
end
subroutine chsign()
C This subroutine changes the sign of all integer arguments passed to it
C and prints the value of all character arguments. It assumes that
C there are at most three integer arguments, followed by the character
C arguments. The way this subroutine is written it has to be compiled
C with the -chararg1 option since it assumes that the lengths of the
C character arguments can be obtained by using GETADR(NARGS+I) where
C NARGS is the number of arguments returned by ARGMNT().
integer*4 val(3)
pointer (stradr, str)
character*(*) str
call argmnt( nargs )
print *, "Number of arguments = ", nargs
j = min(3,nargs)
do 10 i=1,j
call xetarg( i, 4, val(i) )
10 val(i) = - val(i)
C Get the address and the length of the character arguments
do 100 i=j+1, nargs
call getadr( i, stradr )
Page 2
VARARGS(3F)VARARGS(3F)
call getadr( nargs+i-j, lenstr )
write (*,*) str(1:lenstr)
100 continue
call retour(j,4,val(1),4,val(2),4,val(3))
end
Page 3
