This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: C++ PATCH for c++/86981, implement -Wpessimizing-move


On Mon, Aug 20, 2018 at 07:42:10PM -0400, David Malcolm wrote:
> On Mon, 2018-08-20 at 18:54 -0400, Marek Polacek wrote:
> > On Mon, Aug 20, 2018 at 05:18:49PM -0400, David Malcolm wrote:
> > > As of r263675 it's now possible to tell the diagnostics subsystem
> > > that
> > > the warning and note are related by using an auto_diagnostic_group
> > > instance [1], so please can this read:
> > > 
> > >           if (can_do_nrvo_p (arg, functype))
> > >             {
> > >               auto_diagnostic_group d;
> > >               if (warning (OPT_Wpessimizing_move, "moving a local
> > > object "
> > >                            "in a return statement prevents copy
> > > elision"))
> > >                 inform (input_location, "remove %<std::move%>
> > > call");
> > >             }
> > 
> > Sure:
> > 
> > Bootstrapped/regtested on x86_64-linux, ok for trunk?
> 
> Thanks!  The changed part looks good, but I can't really speak to the
> rest of the patch.
> 
> (BTW, could we emit a fix-it hint as part of that "inform" call?  How
> good is the location information at this point?)

Thanks, I can actually use location_of (retval) so that we get:

Wpessimizing-move1.C:43:20: warning: moving a local object in a return statement prevents copy elision [-Wpessimizing-move]
43 |   return std::move (t);
   |          ~~~~~~~~~~^~~

certainly better than what input_location offers.  I'll go with that but
I'll be touching the code today anyway and I'll look into using a fix-it
hint.

Marek


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]