I came across the following piece of code:

#define ERROR  -1
#define STATUS0 0
#define STATUS1 1
#define STATUS2 2

unsigned char foo()
{
  if (/* condition-A */) return ERROR;     /* is that safe? */
  if (/* condition-B */) return STATUS0;
  if (/* condition-C */) return STATUS1;
  return STATUS2;
}

Is '-1 returned as unsigned char' safe in the given context?


Alex Vinokur
     email: alex DOT vinokur AT gmail DOT com
     http://mathforum.org/library/view/10978.html
     http://sourceforge.net/users/alexvn

Alex Vinokur wrote:
> I came across the following piece of code:
>
> #define ERROR  -1
> #define STATUS0 0
> #define STATUS1 1
> #define STATUS2 2
>
> unsigned char foo()
> {
>   if (/* condition-A */) return ERROR;     /* is that safe? */
>   if (/* condition-B */) return STATUS0;
>   if (/* condition-C */) return STATUS1;
>   return STATUS2;
> }
>
> Is '-1 returned as unsigned char' safe in the given context?

In the given (restricted) context, yes, it returns UCHAR_MAX.
[Any value not in the range of an unsigned type is converted
modulo 1+UTYPE_MAX.]

However, in broader contexts, it's not so safe, e.g. ...

  if (foo() == ERROR)

This condition will likely fail even though foo returns ERROR.
This is because the UCHAR_MAX will likely be promoted to a
int, and -1 is not UCHAR_MAX.

-- 
Peter

"Peter Nilsson" <airia@acay.com.au> wrote in message news:1130221108.204867.300010@o13g2000cwo.googlegroups.com...
> Alex Vinokur wrote:
> > I came across the following piece of code:
> >
> > #define ERROR  -1
> > #define STATUS0 0
> > #define STATUS1 1
> > #define STATUS2 2
> >
> > unsigned char foo()
> > {
> >   if (/* condition-A */) return ERROR;     /* is that safe? */
> >   if (/* condition-B */) return STATUS0;
> >   if (/* condition-C */) return STATUS1;
> >   return STATUS2;
> > }
> >
> > Is '-1 returned as unsigned char' safe in the given context?
>
> In the given (restricted) context, yes, it returns UCHAR_MAX.
> [Any value not in the range of an unsigned type is converted
> modulo 1+UTYPE_MAX.]
>
> However, in broader contexts, it's not so safe, e.g. ...
>
>   if (foo() == ERROR)

And in this context?

unsigned char ch1, ch2;
ch1 = foo();
/* Stuff-1 */
if (ch1 == ch2)
{
  /* Stuff-2 */
}

>
> This condition will likely fail even though foo returns ERROR.
> This is because the UCHAR_MAX will likely be promoted to a
> int, and -1 is not UCHAR_MAX.
[snip]


-- 
 Alex Vinokur
     email: alex DOT vinokur AT gmail DOT com
     http://mathforum.org/library/view/10978.html
     http://sourceforge.net/users/alexvn

Alex Vinokur wrote:
> And in this context?
>
> unsigned char ch1, ch2;
> ch1 = foo();
> /* Stuff-1 */
> if (ch1 == ch2)
> {
>   /* Stuff-2 */
> }

and in this context ch2 is undefined :-p

<makc.the.great@gmail.com> wrote in message news:1130224992.505138.43120@g47g2000cwa.googlegroups.com...
>
> Alex Vinokur wrote:
> > And in this context?
> >
> > unsigned char ch1, ch2;
> > ch1 = foo();
> > /* Stuff-1 */
> > if (ch1 == ch2)
> > {
> >   /* Stuff-2 */
> > }
>
> and in this context ch2 is undefined :-p
>

Stuff-1 can change ch2.


-- 
 Alex Vinokur
     email: alex DOT vinokur AT gmail DOT com
     http://mathforum.org/library/view/10978.html
     http://sourceforge.net/users/alexvn

Alex Vinokur wrote:
> I came across the following piece of code:
> 
> #define ERROR  -1
> #define STATUS0 0
> #define STATUS1 1
> #define STATUS2 2
> 
> unsigned char foo()
> {
>   if (/* condition-A */) return ERROR;     /* is that safe? */
[...]
> Is '-1 returned as unsigned char' safe in the given context?

[comp.lang.c answer]
Yes, it is safe in the sense that UCHAR_MAX will be returned.
No, it is not safe, because 'ERROR' invades the namespace reserved to 
the implementation.

As a general rule, crossposting to comp.lang.c and comp.lang.c++ is a 
poor idea.  Not only are these different languages, but when the syntax 
is the same in each language, the semantics may differ, and when the 
syntax and semantics agree, the accepted best practice may differ.

Alex Vinokur wrote:
> > > unsigned char ch1, ch2;
> > > ch1 = foo();
> > > /* Stuff-1 */
> > > if (ch1 == ch2)
> > > {
> > >   /* Stuff-2 */
> > > }
> >
> > and in this context ch2 is undefined :-p
> >
>
> Stuff-1 can change ch2.

A comment can change variable? Now that's something new :-p''

Martin Ambuhl wrote:
> Alex Vinokur wrote:
> > I came across the following piece of code:
> >
> > #define ERROR  -1
> > #define STATUS0 0
> > #define STATUS1 1
> > #define STATUS2 2
> >
> > unsigned char foo()
> > {
> >   if (/* condition-A */) return ERROR;     /* is that safe? */
> [...]
> > Is '-1 returned as unsigned char' safe in the given context?
>
> [comp.lang.c answer]
> Yes, it is safe in the sense that UCHAR_MAX will be returned.
> No, it is not safe, because 'ERROR' invades the namespace reserved
> to the implementation.

True, although EXXXX identifiers are only reserved if the code
includes <errno.h>. That said, it is best to avoid defining such
identifiers at all, in case the code ever gets included by other
code that might use <errno.h>.

-- 
Peter