Re: coding problem using salford FTN95
From: Richard E Maine (nospam_at_see.signature)
Date: 11/03/04
- Next message: Kamaraju Kusumanchi: "Re: GNU Fortran 95: Opinions?"
- Previous message: Dr. Flonkenstein: "Re: digest 2453311"
- In reply to: bab: "coding problem using salford FTN95"
- Next in thread: jan van oosterwijk: "Re: coding problem using salford FTN95"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 02 Nov 2004 16:33:10 -0800
"bab" <bab@bab.bab> writes:
> The problem is that he cannot make it calculate for values less than 0 but
> above it is ok
You don't say what "cannot make it calculate" means. What does he
expect to happen and what actually does happen? I see nothing in the
Fortran code that should particularly have a problem with values less
than zero.
However, values less than zero make no physical sense here. This has
nothing to do with programming. Just about the only sensible thing
to do with a value less than zero for such things as a diameter is to
print an error message. I would say that "not calculating" for
values less than zero is exactly what I would want the program to do.
I'd be more concerned that it does calculate for such numbers when
ideally it shouldn't.
I do see one coding problem with the program. It wouldn't particularly
cause problems for values less than zero, but it would cause problems
for values of length that were not integers. These problems would be
most obvious for values of length less than 1, so perhaps you meant
less than 1 instead of less than zero. There is actually a problem
for any non-integer value, but the problem is most obvious when
the value is less than 1 (the result would be zero).
Namely, the code does not declare the types of any of its variables.
Fortran lets you get by with that, but I *STRONGLY* recommend
declaring the types of all variables.
Firstly, I think it really important that the programmer think about
the type of every variable. It is a very important subject in
programming; if you don't think at least a little about it, you will
have no end of problems. Declaring the type tends to make you think
at least a little about it.
Secondly, if you don't declare the types of your variables, and
don't think about the subject, you are likely to find that they have
the wrong type, as presumably happened here. All the variables
other then length in this code will be implicitly declared as
real, which is appropriate, but length will be implicitly
declared as integer, which is not a good choice.
At the very least, your friend needs to add a declaration such as
real :: length
to the code. I recommend also declaring all the other variables
to be real and adding a statement
implicit none
right after the program statement. This statement will turn off
the compilers implicit typing so that you will get an error message
if you accidentally fail to declare the type of a variable. This
helps avoid the kind of problem that this code has.
> Program
Not really related, but I'm surprised that the above line worked.
It isn't standard Fortran. The program statement is supposed to
have a program name. Doesn't much matter what the name is, but
there is suposed to be one. Something like (choose any better
name if you like, but this suffices for illustration)
Program my_program
-- Richard Maine | Good judgment comes from experience; email: my first.last at org.domain | experience comes from bad judgment. org: nasa, domain: gov | -- Mark Twain
- Next message: Kamaraju Kusumanchi: "Re: GNU Fortran 95: Opinions?"
- Previous message: Dr. Flonkenstein: "Re: digest 2453311"
- In reply to: bab: "coding problem using salford FTN95"
- Next in thread: jan van oosterwijk: "Re: coding problem using salford FTN95"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
