Re: [RFA 1/2] Fix two regressions in scalar printing

[Eli Zaretskii <eliz at gnu dot org>]

> From: Pedro Alves <palves@redhat.com>
> Date: Fri, 14 Jul 2017 15:56:27 +0100
> 
> The manual says:
> 
>  @item d
>  Print as integer in signed decimal.
> 
>  @item u
>  Print as integer in unsigned decimal.
> 
> And I see this with a gdb from before the recent print_scalar_formatted
> changes:
> 
>  (gdb) p /d (unsigned long long) -1
>  $1 = -1
> 
> while we see this with either current master, or your patch:
> 
>  (gdb) p /d (unsigned long long) -1
>  $1 = 18446744073709551615
> 
> which also doesn't look right to me.
> 
> And here:
> 
>  (gdb) p /d (unsigned) -1
>  $2 = 4294967295
> 
> I'd expect "-1", but we don't get it with any gdb version (before
> original print_scalar_formatted changes, or current master, or
> your current patch), which also looks like a bug to me.

I think I agree: we should produce what the format says, not what the
cast says.