Re: [PATCH 07/43] Move get_object_address to dwarf_expr_context

[Zoran Zaric via Gdb-patches <gdb-patches@sourceware.org>]

On 4/27/21 4:12 AM, Simon Marchi wrote:
> [CAUTION: External Email]
> 
>> @@ -204,7 +207,12 @@ struct dwarf_expr_context
>>                                           int deref_size) = 0;
>>
>>     /* Return the `object address' for DW_OP_push_object_address.  */
>> -  virtual CORE_ADDR get_object_address () = 0;
>> +  virtual CORE_ADDR get_object_address ()
>> +  {
>> +    if (obj_address == 0)
>> +      error (_("Location address is not set."));
>> +    return obj_address;
> 
> Instead of having obj_address == 0 as a special case, I'd suggest making
> it a gdb::optional maybe.  To make sure that nobody accesses it
> directly, other than through the getter, you could even apply the same
> pattern of wrapping it that I suggested for the frame and per_cu.
> 
> Should the error thrown here be of the same type as the error thrown for
> if the frame or per_cu is not there (NOT_AVAILABLE_ERROR currently, but
> perhaps subject to change)?
> 
> Simon
> 

This method was just copied from the previous implementation and inlined 
in the later patches.

Zoran