portability and return

**Walter Roberson** · Jun 27 '08, 07:37 PM

Re: portability and return

In article <2YnWj.20544$5b 3.14006@trnddc0 5>,
Bill Cunningham <nospam@nspam.c omwrote:

I have heard that return -1 is not portable. So what would be the answer
>to portability in this case? What about exit(0) and exit (-1) or exit (1)?
>Or would it be best to stick with C's macros, hence:

>exit(EXIT_FAIL URE)
>exit (EXIT_SUCCESS)

It would be best to stick with C's macros. The results are implementation
defined if you use any value other than 0 or one of the macros.
In particular, exit(1) is *not* something that C assigns any meaning to.
--
"What is important, then, is not that the critic should possess a
correct abstract definition of beauty for the intellect, but a
certain kind of temperament, the power of being deeply moved by
the presence of beautiful objects." -- Walter Pater

**Peter Nilsson** · Jun 27 '08, 07:37 PM

Re: portability and return

Bill Cunningham wrote:

I have heard that return -1 is not portable.

Presumably you're talking about the return value from main()?

So what would be the answer to portability in this case?
What about exit(0) and exit (-1) or exit (1)?

exit(0) is fine.

Or would it be best to stick with C's macros, hence:
>
exit(EXIT_FAILU RE)
exit (EXIT_SUCCESS)

Those are good too.

--
Peter

**Bill Cunningham** · Jun 27 '08, 07:37 PM

Re: portability and return

"Peter Nilsson" <airia@acay.com .auwrote in message
news:3222fc05-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...

Bill Cunningham wrote:

>I have heard that return -1 is not portable.

>
Presumably you're talking about the return value from main()?

Yes. Would return -1 from a function other than main be ok? I'm guessing
no but I thought I'd ask.

Bill

**Spiros Bousbouras** · Jun 27 '08, 07:37 PM

Re: portability and return

On 13 May, 22:57, "Bill Cunningham" <nos...@nspam.c omwrote:

"Peter Nilsson" <ai...@acay.com .auwrote in message
>
news:3222fc05-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...
>

Bill Cunningham wrote:

I have heard that return -1 is not portable.

>

Presumably you're talking about the return value from main()?

>
Yes. Would return -1 from a function other than main be ok? I'm guessing
no but I thought I'd ask.

exit(-1) is not portable no matter which function
calls it.

**Harald van =?UTF-8?b?RMSzaw==?=** · Jun 27 '08, 07:37 PM

Re: portability and return

On Tue, 13 May 2008 21:57:23 +0000, Bill Cunningham wrote:

Yes. Would return -1 from a function other than main be ok? I'm
guessing
no but I thought I'd ask.

Are you asking if it's okay to return -1 in such cases as calling
add(2, -3), defined as

int add(int a, int b) {
return a + b;
}

? What else would you want to make this function return?

**Walter Roberson** · Jun 27 '08, 07:37 PM

Re: portability and return

In article <79oWj.7860$Uz2 .2677@trnddc06> ,
Bill Cunningham <nospam@nspam.c omwrote:

Yes. Would return -1 from a function other than main be ok? I'm guessing
>no but I thought I'd ask.

For any function other than main, it is fine to return to its
caller any value that lies within the range of the type of the return
value.

But, just in case you meant something different: if you are calling
exit() from *anywhere* in the program, you should restrict the
exit value to 0 or one of the two status macros.
--
"Whenever there is a hard job to be done I assign it to a lazy
man; he is sure to find an easy way of doing it."
-- Walter Chrysler

**Bill Cunningham** · Jun 27 '08, 07:37 PM

Re: portability and return

"Harald van D?k" <truedfx@gmail. comwrote in Are you asking if it's okay
to return -1 in such cases as calling

add(2, -3), defined as
>
int add(int a, int b) {
return a + b;
}
>
? What else would you want to make this function return?

No no. Not in this case. Of course return as you specified is correct I
understand. I mean to simply exit with say and error.

if ( argc != 5)
{ fprintf(stderr, "error\n");
return -1; /* exit(-1) is wrong I guess the proper return would be */
exit(EXIT_FAILU RE);
}

Bill

**jacob navia** · Jun 27 '08, 07:37 PM

Re: portability and return

Bill Cunningham wrote:

"Peter Nilsson" <airia@acay.com .auwrote in message
news:3222fc05-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...

>Bill Cunningham wrote:

>>I have heard that return -1 is not portable.

>Presumably you're talking about the return value from main()?

>
Yes. Would return -1 from a function other than main be ok? I'm guessing
no but I thought I'd ask.
>
Bill
>
>

Obviously returning -1 from ANY function (including main)
is portable and well defined by the language.

Now, returning -1 from main could be interpreted by the OS
(when there is one) in different ways and could be non
portable in the sense that it could mean different things in
different OSes.

But this is after main returns, so it is no longer a C problem.

By the way, I find all this discussion completely stupid sorry.

What do you have against -1???

It is just as good a number as 1, or 4477665 for that matter.

--
jacob navia
jacob at jacob point remcomp point fr
logiciels/informatique

403 Forbidden

http://www.cs.virginia.edu/~lcc-win32

**Richard Tobin** · Jun 27 '08, 07:37 PM

Re: portability and return

In article <79oWj.7860$Uz2 .2677@trnddc06> ,
Bill Cunningham <nospam@nspam.c omwrote:

Yes. Would return -1 from a function other than main be ok?

Yes of course. You can return any int from a function declared as
returning an int.

There's no problem returning -1 from main() either, as far as the C
program is concerned. All that's undefined is how the "host
environment" - the shell or operating system or parent program -
interprets it. 0 (or EXIT_SUCCESS) indicates success. EXIT_FAILURE
indicates failure. What anything else means depends on your system.

Most of my programs return 0, 1, or sometimes a larger value, which
can be interpreted reasonably in the unix environments I intend them
for. But you can perfectly well run them in other environments; you
just have to ensure that you run them in such a way that the return
value is not misinterpreted. I suppose it's possible that somewhere
there's an operating system that always shuts down completely if any
C program calls exit(1), but the solution to that is to get a different
operating system.

-- Richard
--
:wq

**Bill Cunningham** · Jun 27 '08, 07:37 PM

Re: portability and return

"Richard Tobin" <richard@cogsci .ed.ac.ukwrote in message
news:g0d3t6$26u i$1@pc-news.cogsci.ed. ac.uk...

There's no problem returning -1 from main() either, as far as the C
program is concerned. All that's undefined is how the "host
environment" - the shell or operating system or parent program -
interprets it. 0 (or EXIT_SUCCESS) indicates success. EXIT_FAILURE
indicates failure. What anything else means depends on your system.

[snip]

That's what I am concerned with. How the host system will react. What I
am really concerned with is what the standard says would run best on any
environment. I write small C programs in linux and I want to be able to take
the same source code to my windows system and compile it there and get the
same results. I am trying to get into the habit of treating all code as
development code so it would work in all host environments.

Bill

**pete** · Jun 27 '08, 07:37 PM

Re: portability and return

Bill Cunningham wrote:

"Peter Nilsson" <airia@acay.com .auwrote in message
news:3222fc05-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...

>Bill Cunningham wrote:

>>I have heard that return -1 is not portable.

>Presumably you're talking about the return value from main()?

>
Yes. Would return -1 from a function other than main be ok?
I'm guessing no but I thought I'd ask.

You can write another function that returns -1 if you want to.
There's no problem with that.
A lot of functions return EOF.
Any negative int, is an allowable value
for an implementation to use to define EOF.

The 3 portable return values from main are:
0
EXIT_SUCCESS
EXIT_FAILURE

--
pete

**Jens Thoms Toerring** · Jun 27 '08, 07:37 PM

Re: portability and return

jacob navia <jacob@nospam.o rgwrote:

Bill Cunningham wrote:

"Peter Nilsson" <airia@acay.com .auwrote in message
news:3222fc05-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...

Bill Cunningham wrote:
>I have heard that return -1 is not portable.
Presumably you're talking about the return value from main()?

Yes. Would return -1 from a function other than main be ok? I'm guessing
no but I thought I'd ask.

Bill

Obviously returning -1 from ANY function (including main)
is portable and well defined by the language.

Now, returning -1 from main could be interpreted by the OS
(when there is one) in different ways and could be non
portable in the sense that it could mean different things in
different OSes.

But this is after main returns, so it is no longer a C problem.

By the way, I find all this discussion completely stupid sorry.

What do you have against -1???

It is just as good a number as 1, or 4477665 for that matter.

Jacob, I am rather sure you know better, so just for the OP. If
you want to return a value from main() that indicates success
to the system that invoked the program (and that holds for all
systems) then return either 0 or EXIT_SUCCESS. If you want to
return a value that indicates failure return EXIT_FAILURE.
That's what the C standard requires. It even might mean that
when you return 0 (or EXIT_SUCCESS) the calling system will
receive a value of 42 if it expects 42 to mean success (and
-53 if you return EXIT_FAILURE and on that system -53 indi-
cates failure).

With everything else you don't have such a guarantee. That
makes it not wrong per se to return something else but it
restricts the usability of the return value to those systems
that you know what return value they expect (and e.g. 4477665
won't be anything useful for UNIX systems where the return
value has to fit into 8 bits).

Regards, Jens
--
\ Jens Thoms Toerring ___ jt@toerring.de
\______________ ____________ http://toerring.de

**Eric Sosman** · Jun 27 '08, 07:37 PM

Re: portability and return

jacob navia wrote:

Bill Cunningham wrote:

>"Peter Nilsson" <airia@acay.com .auwrote in message
>news:3222fc0 5-7aee-4eb3-b874-eb55981751dd@v2 6g2000prm.googl egroups.com...

>>Bill Cunningham wrote:
>>>I have heard that return -1 is not portable.
>>Presumably you're talking about the return value from main()?

>>
> Yes. Would return -1 from a function other than main be ok? I'm
>guessing no but I thought I'd ask.
>>
>Bill
>>
>>

>
Obviously returning -1 from ANY function (including main)
is portable and well defined by the language.
>
Now, returning -1 from main could be interpreted by the OS
(when there is one) in different ways and could be non
portable in the sense that it could mean different things in
different OSes.
>
But this is after main returns, so it is no longer a C problem.
>
By the way, I find all this discussion completely stupid sorry.
>
What do you have against -1???
>
It is just as good a number as 1, or 4477665 for that matter.

A few years ago somebody asked about systems where a
termination status of -1 would survive its journey to the
environment and be meaningful when it got there. As far
as I can recall, nobody could name such a system. The
somewhat bemused question, of course, was: If there aren't
any systems where -1 makes sense (or if they're so rare
that nobody can think of any), why is -1 so often used
as a termination status?

(Open)VMS was a near-miss, though, since the value would
make it unscathed all the way out to the invoker. However,
VMS' "condition codes" contain various sub-fields in their
32 bits, and the value -1 would set the "severity" sub-field
to a reserved-to-HP value, would set a bit that inhibits the
display of a termination message, and would set three "must
be zero" bits non-zero. So: survivable, but not meaningful.

--
Eric.Sosman@sun .com

**Bill Cunningham** · Jun 27 '08, 07:37 PM

Re: portability and return

"Jens Thoms Toerring" <jt@toerring.de wrote in message
news:68uj7rF2ua tl5U1@mid.uni-berlin.de...

Jacob, I am rather sure you know better, so just for the OP. If
you want to return a value from main() that indicates success
to the system that invoked the program (and that holds for all
systems) then return either 0 or EXIT_SUCCESS. If you want to
return a value that indicates failure return EXIT_FAILURE.
That's what the C standard requires. It even might mean that
when you return 0 (or EXIT_SUCCESS) the calling system will
receive a value of 42 if it expects 42 to mean success (and
-53 if you return EXIT_FAILURE and on that system -53 indi-
cates failure).
>
With everything else you don't have such a guarantee. That
makes it not wrong per se to return something else but it
restricts the usability of the return value to those systems
that you know what return value they expect (and e.g. 4477665
won't be anything useful for UNIX systems where the return
value has to fit into 8 bits).
>
Regards, Jens

Thanks very much. That answers my question completely.

Bill

portability and return

portability and return

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment