gcc not warning on all occurences of operations with undefined results?

gcc not warning on all occurences of operations with undefined results?

[Daniel Kabs]

Hello there,

gcc 3.3.5 issues warnings on lines like

  a[i] = i++;
  i = i++;
  *p++ = func(*p);

e.g.
"test_increment.c:19: warning: operation on `i' may be undefined"

I take it, this is because of the C standard saying, that if I have an 
expression and somewhere in that expression, I use an increment (decrement) 
operator on a variable, I can not mention that variable elsewhere in the 
expression.

The warnings are very helpful in finding badly written expressions.

But why doesn't gcc warn about this code

  s.a = s.a++;
  p->a = p->a++;

IMHO, these lines also infringe upon the above mentioned rule or did I miss 
something important? 

Cheers
Daniel 

Re: gcc not warning on all occurences of operations with undefined results?

[John Love-Jensen]

Hi Daniel,

> IMHO, these lines also infringe upon the above mentioned rule or did I miss
> something important?

The compiler *CAN* warn about those kinds of infringements.

The compiler does *NOT* *HAVE* to warn about those kinds of infringements.

GCC 3.3.5 detects the simple case.  It does not (I presume) detect the
indirect case that you provided.

In some situations it may be very hard to detect the indirect case during
compile time, and it could be very expensive to check for it at run time.

HTH,
--Eljay

Re: gcc not warning on all occurences of operations with undefined results?

[Ladislav Mecir]

Daniel Kabs napsal(a):

>Hello there,
>
>gcc 3.3.5 issues warnings on lines like
>
>  a[i] = i++;
>  i = i++;
>  *p++ = func(*p);
>
>e.g.
>"test_increment.c:19: warning: operation on `i' may be undefined"
>
>I take it, this is because of the C standard saying, that if I have an 
>expression and somewhere in that expression, I use an increment (decrement) 
>operator on a variable, I can not mention that variable elsewhere in the 
>expression.
>
>  
>
I think, that this is wrong. The exact rule should be more like "...if 
the result of the operation is not defined by standard..."

>The warnings are very helpful in finding badly written expressions.
>
>But why doesn't gcc warn about this code
>
>  s.a = s.a++;
>  p->a = p->a++;
>  
>
because in both these cases the result *is* defined

>IMHO, these lines also infringe upon the above mentioned rule or did I miss 
>something important? 
>
>Cheers
>Daniel 
>  
>
-L

Re: gcc not warning on all occurences of operations with undefined results?

[Daniel Kabs]

Hello Ladislav!

On Tuesday 15 November 2005 21:33, Ladislav Mecir wrote:
> >But why doesn't gcc warn about this code
> >
> >  s.a = s.a++;
> >  p->a = p->a++;
>
> because in both these cases the result *is* defined

I do not understand, why are those operations defined, but

i = i++;

is not? I fail to see the relevant difference in these assigments.

Thanks for your help,

Cheers
Daniel