On Tuesday, August 19, 2008 07:55:37 pm Tom Tromey wrote:
> +  /* Flags about this type.  These fields appear at this location
> +     because they packs nicely here.  See the TYPE_* macros for
> +     documentation about these fields.  */
> +
> +  unsigned int flag_unsigned : 1;
> +  unsigned int flag_nosign : 1;
> +  unsigned int flag_stub : 1;
> +  unsigned int flag_target_stub : 1;
> +  unsigned int flag_static : 1;
> +  unsigned int flag_prototyped : 1;
> +  unsigned int flag_incomplete : 1;
> +  unsigned int flag_varargs : 1;
> +  unsigned int flag_vector : 1;
> +  unsigned int flag_stub_supported : 1;
> +  unsigned int flag_nottext : 1;
> +  unsigned int flag_fixed_instance : 1;

Hi Tom,

This is quite an old change but while debugging gdb I noticed that vector 
types do have a strange bit set into their instance_flags and this seems to go 
back to this patch.
The snippet above introduces the flag_nottext as a bitfield member of the type 
struct while gdbtypes.c:make_vector_type still sets that bit into the 
instance_flags.

The nottext flag is set for the element types of vectors 
(gdbtypes.c:make_vector_type) and for the builtin_int8/builtin_int8 types. The 
flag is read from the c-valprint.c:c_textual_element_type function that 
determines whether arrays of chars should be printed as strings or not.  So, I 
guess that prior to this patch char vectors were printed just like integer 
vectors - plain data. One approach to restore that functionality would be to 
move the nottext flag into to the instance_flags of the type. Attached is an 
untested patch of what I have in mind. Comments are welcome.

This also renders my previous attempt to fix the printing of character vectors 
(http://sourceware.org/ml/gdb-patches/2010-06/msg00573.html) obsolete.

Regards
Ken Werner