Re: [C++ Patch] PR 63985

[Paolo Carlini <paolo dot carlini at oracle dot com>]

Hi,

On 12/23/2014 07:13 PM, Jason Merrill wrote:
> On 12/23/2014 11:30 AM, Paolo Carlini wrote:
> >         if (maybe_range_for_decl)
> > -    *maybe_range_for_decl = error_mark_node;
> > +    {
> > +      *maybe_range_for_decl = error_mark_node;
> > +      if (*equal_loc == UNKNOWN_LOCATION)
> > +        tmp_equal_loc = token->location;
>
> Even though the range-for case is the only place we care about the 
> initializer location, I'd rather set *equal_loc whenever equal_loc is 
> non-null.  We can also use the local initializer location in place of 
> initializer_start_token->location.
>
> And let's call it init_loc rather than equal_loc.
Ok. I think the below is closer, the only detail I want to mention is 
that it still checks *init_loc == UNKNOWN_LOCATION before assigning, 
because the location of the error doesn't change if more than one 
initializer is present. Likewise for the comma.

Thanks,
Paolo.

///////////////////////

Attachment: patch_63985_5
Description: Text document