Re: [PATCH][gdb/ada] Handle artificial local symbols

[Tom de Vries <tdevries@suse.de>]

On 9/17/21 9:42 PM, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
> Tom> Fix this instead by marking the symbol as artificial, and:
> Tom> - ignoring such symbols in ada_resolve_variable, which fixes the FAIL
> Tom> - ignoring such ada symbols in do_print_variable_and_value, which prevents
> Tom>   them from showing up in "info locals"
> 
> Tom> Note that a fix for the latter was submitted here (
> Tom> https://sourceware.org/pipermail/gdb-patches/2008-January/054994.html ), and
> Tom> this patch borrows from it.
> 
> Thanks for doing this.
> 
> Tom> +      /* Handle DW_AT_artificial.  */
> Tom> +      attr = dwarf2_attr (die, DW_AT_artificial, cu);
> Tom> +      if (attr != nullptr)
> Tom> +	sym->artificial = 1;
> 
> This should also check attr->as_boolean.
> Probably also use '= true', though
> 
>    sym->artificial = attr->as_boolean ()
> 
> would also be fine.
> 

Ack.

> Tom> +  if (language_def (sym->language ())->symbol_printing_suppressed (sym))
> Tom> +    return;
> 
> Seems fine though I also wonder if it would be better to just always
> suppress artificial variables here.
> 

There may be trouble with other languages though.  So for now I'm going
ahead with this approach.

> Tom> +  /* Whether this symbol is artificial.  */
> Tom> +
> Tom> +  unsigned int artificial : 1;
> 
> I think making this bool would be better.

Hmm, the struct is advertised as space-critical, and a 1-bit bitfield is
smaller than an 8-bit bool.  So I wonder why you prefer bool here.

Thanks,
- Tom