simonnet@geocub.greco-prog.fr (09/10/89)
1 = 3 ......
For some machines (HP1000 for example) 1 = 3
program toto
call add (1)
write (1,*) 'One = ?', 1
end
*--------------------------------------*
subroutine add (a)
integer *2 a
a = a + 2
return
end
Result is : One = ? 3 !!!!!brainerd@unmvax.unm.edu (Walt Brainerd) (09/11/89)
In article <1349@geocub.greco-prog.fr>, simonnet@geocub.greco-prog.fr writes: > For some machines (HP1000 for example) 1 = 3 > > program toto > > call add (1) > write (1,*) 'One = ?', 1 > end > > *--------------------------------------* > subroutine add (a) > > integer *2 a > > a = a + 2 > return > > end > As I am sure many will point out, this program is illegal. The standard (p. 15-16, lines 14-20): "Actual arguments may be constants ... if and only if the associated dummy argument is a variable that is not defined during execution of the referenced external procedure." And this is one of those cases which is not likely to be caught even by a system that does quite a bit of error checking, since the error is "across" two program units. Of course, the program is also illegal because of the presence of integer *2 -- Walt Brainerd Unicomp, Inc. brainerd@unmvax.cs.unm.edu 2002 Quail Run Dr. NE Albuquerque, NM 87122 505/275-0800
jerry@violet.berkeley.edu ( Jerry Berkman ) (09/12/89)
In article <1349@geocub.greco-prog.fr> simonnet@goofi.UUCP (Simonnet Thierry) writes: > > 1 = 3 ...... > > > For some machines (HP1000 for example) 1 = 3 > > > program toto > > call add (1) > write (1,*) 'One = ?', 1 > end > > *--------------------------------------* > subroutine add (a) > > integer *2 a > > a = a + 2 > return > > end > >Result is : One = ? 3 !!!!! As Walt Brainard points out, changing the argument when the argument is a constant is non-standard, as is integer*2. Even on those systems which have the "integer*2" extension, it's doubtful that they all do the same thing for passing an integer constant like 1 to an integer*2 argument. Testing shows no consistent pattern: On Cray UNICOS (cft & cft77): output is "One = ?1" On IBM 3090: error: unit number out of range (using "1" as unit #!) On IBM 3090 writing to unit * instead of unit 1: output is: "One = ?131073" On sun 3/50: segmentation fault & output is 641 chars. of garbage On sun 3/50 without "*2": no seg. fault, output is 50 garbage chars. On VAX with BSD f77: BSS error, no output. (constants are stored in read only memory). On VAX with DEC Fortran: BSS error, no output. - Jerry Berkman, U.C.Berkeley, jerry@violet.berkeley.edu
khb@road.Sun.COM (Keith Bierman - Advanced Languages - Floating Point Group ) (09/12/89)
In article <375@unmvax.unm.edu> brainerd@unmvax.unm.edu (Walt Brainerd) writes: > discussion of how 1=3 >And this is one of those cases which is not likely to be caught even >by a system that does quite a bit of error checking, since the error >is "across" two program units. Lahey's PC fortran has a mode which makes this code work as one expects (viz. 1=1). Sun Fortran seems to protect 1 as well. Keith H. Bierman |*My thoughts are my own. !! kbierman@sun.com It's Not My Fault | MTS --Only my work belongs to Sun* I Voted for Bill & | Advanced Languages/Floating Point Group Opus | "When the going gets Weird .. the Weird turn PRO"