[C++ Patch] Improve build_new_op_1, cp_build_indirect_ref_1, and cp_build_modify_expr locations

[Jason Merrill]

On 11/22/19 10:43 AM, Paolo Carlini wrote:
> Hi,
> 
> I would say most of the changes are straightforward or mechanical. 
> Essentially, for build_new_op_1 and cp_build_modify_expr I'm simply 
> consistently using the available location argument; for 
> cp_build_indirect_ref_1 I'm adding the parameter but then using it in a 
> completely straightforward way. Minor nit: I wondered for a while if 
> cp_build_modify_expr should use cp_expr_loc_or_loc more - normally the 
> passed loc points to the '=' - but eventually, given the actual texts of 
> the messages, I used it only in one place, for "void value not ignored 
> as it ought to be" which is mostly about the type of 'rhs'. All the 
> other messages in one way or the other talk about both sides (the 
> primary clang caret appears to agree).
> 
> Tested x86_64-linux.
> 
> Thanks, Paolo.
> 
> //////////////////////////////
> 
OK.

Jason