Re: output-radix breaks char output

[Pedro Alves <pedro at codesourcery dot com>]

On Tuesday 08 November 2011 04:39:08, Mike Frysinger wrote:
> am i doing something wrong here ?  seems like setting output-radix to 16 
> breaks x/c output.
> 
> $ echo 'char f[]="abc";main(){}' | gcc -x c -g - -o a.out
> $ gdb ./a.out
> (gdb) b main
> (gdb) r
> (gdb) set output-radix 10
> Output radix now set to decimal 10, hex a, octal 12.
> (gdb) x/4c f
> 0x601018 <f>:   97 'a'  98 'b'  99 'c'  0 '\000'
> (gdb) set output-radix 16
> Output radix now set to decimal 16, hex 10, octal 20.
> (gdb) x/4c f
> 0x601018 <f>:   0x61    0x62    0x63    0x0
> 
> i can understand the 0x61/97 change, but imo, the latter output shouldn't be 
> missing the 'a' bits.

I agree.

static void
set_output_radix_1 (int from_tty, unsigned radix)
{
  /* Validate the radix and disallow ones that we aren't prepared to
     handle correctly, leaving the radix unchanged.  */
  switch (radix)
    {
    case 16:
      user_print_options.output_format = 'x';	/* hex */
      break;
    case 10:
      user_print_options.output_format = 0;	/* decimal */
      break;
    case 8:
      user_print_options.output_format = 'o';	/* octal */
      break;
    default:

So radix==10 means "no format", and other radices work as if
the user overrode the format whenever printing, basically
the same as:

(top-gdb) p **argv
$3 = 47 '/'

GDB knew that was a char from the type.

(top-gdb) p /d **argv
$4 = 47

... when overridden as decimal number, the '/' is gone.

(top-gdb) x /d *argv
0x7fffffffe30a: 47

... same with `x' (the machinery that decides this is basically the same).

(top-gdb) p /x **argv
$5 = 0x2f
(top-gdb) p /o **argv
$6 = 057
(top-gdb) p /c **argv
$7 = 47 '/'

... and other formats too.

-- 
Pedro Alves