[Bug c/15657] New: -wall incorrectly diagnoses scanf("%a")

[Bug c/15657] New: -wall incorrectly diagnoses scanf("%a")

[paulmoore100 at hotmail dot com]

3.3.1 (suse 9) 3.2.2 (red hat9) 
#include<stdio.h>
void f()
{
 char *s=0;
 scanf("%a", &s);
}

cc -Wall on this incorrectly produces "float format, pointer arg(arg2)"
it should say "invalid or incomplete format" or something like that

-- 
           Summary: -wall incorrectly diagnoses scanf("%a")
           Product: gcc
           Version: 3.3.1
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: paulmoore100 at hotmail dot com
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-26 02:38 -------
Confirmed, the warning should be better.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |diagnostic
   Last reconfirmed|0000-00-00 00:00:00         |2004-05-26 02:38:31
               date|                            |
            Summary|-wall incorrectly diagnoses |-Wformat diagnostic
                   |scanf("%a")                 |confusing


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[jsm at polyomino dot org dot uk]

------- Additional Comments From jsm at polyomino dot org dot uk  2004-05-26 09:00 -------
Subject: Re:  New: -wall incorrectly diagnoses scanf("%a")

On Wed, 26 May 2004, paulmoore100 at hotmail dot com wrote:

> 3.3.1 (suse 9) 3.2.2 (red hat9) 
> #include<stdio.h>
> void f()
> {
>  char *s=0;
>  scanf("%a", &s);
> }
> 
> cc -Wall on this incorrectly produces "float format, pointer arg(arg2)"
> it should say "invalid or incomplete format" or something like that

What do you think is wrong with the existing message?  %a is one of the
eight formats for a float (along with %e, %f, %g, %A, %E, %F, %G); C99
subclause 7.19.6.2 paragraphs 12 and 14.  There is a glibc extension which
conflicts with this, meaning that glibc doesn't conform to C99 in this
area (and it doesn't conform to C90 for scanf either), but it only applies
when %a is immediately followed by 's', 'S' or '['.  Even in those cases
it is nonconforming to accept that extension; and, therefore, in C99 mode
GCC does not recognise that glibc extension at all in format checking, and
hopefully glibc will be fixed in future not to recognise it either in
strict C99 mode.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-26 11:43 -------
I think the confusing part to me is that the really does not say that the format is different from what is 
supplied as an arg.

Maybe the following should be the error.
float format supplied but got a pointer argument (arg2).

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[paulmoore100 at hotmail dot com]

------- Additional Comments From paulmoore100 at hotmail dot com  2004-05-26 17:17 -------
is %a valid for scanf? I dont think it is, f is used for float on input a is 
used on oputput

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-26 17:28 -------
That is because glibc has an extension before C99 came out.  see <http://gnu.gnusoft.net/software/
libc/CONFORMANCE>.

C99 added %a as another scanf format specifier for floating point
values.  This conflicts with the glibc extension where %as, %a[ and
%aS mean to allocate the string for the data read.  A strictly
conforming C99 program using %as, %a[ or %aS in a scanf format string
will misbehave under glibc.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[paulmoore100 at hotmail dot com]

------- Additional Comments From paulmoore100 at hotmail dot com  2004-05-26 18:57 -------
THen maybe the message is correct - except it should say 
float format , char * argument
it would be nice if it said "did you mean %as?" but thats probably pushing it 
too far


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[jsm at polyomino dot org dot uk]

------- Additional Comments From jsm at polyomino dot org dot uk  2004-05-26 22:18 -------
Subject: Re:  -Wformat diagnostic confusing

On Wed, 26 May 2004, paulmoore100 at hotmail dot com wrote:

> THen maybe the message is correct - except it should say 
> float format , char * argument

That messages could name the types involved better is bug 1027.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657

[Bug c/15657] -Wformat diagnostic confusing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-26 23:04 -------
So closing as a dup of bug 1027.

*** This bug has been marked as a duplicate of 1027 ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15657