Re: C++ PATCH: fix PC 10903

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Tuesday, Jul 15, 2003, at 21:53 US/Eastern, Phil Edwards wrote:

> On Wed, Jul 16, 2003 at 03:42:52AM +0200, Gabriel Dos Reis wrote:
> > *** cp/ChangeLog	16 Jul 2003 01:41:40 -0000
> > --- cp/ChangeLog	16 Jul 2003 00:09:41 -0000	1.3526
> > ***************
> > *** 1,9 ****
> > - 2003-07-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
> > -
> > - 	PR c++/10903
> > - 	* pt.c (convert_nontype_argument): Fix thinko in diagnostic.
> > - 	Improve.
> > -
>
> Both the changelog entry and the testcase show up as removals rather
> than additions.  Could you check that please, just to be safe?

The patch was reversed for some reason so you are reading the patch the 
opposite way.
> > *************** convert_nontype_argument (tree type, tre
> > *** 3085,3098 ****
> >       }
> >     else
> >       {
> > !       if (TYPE_P (expr))
> > !         error ("type '%T' cannot be used as a value for a non-type "
> > !                "template-parameter", expr);
> > !       else if (DECL_P (expr))
> > !         error ("invalid use of '%D' as a non-type 
> > template-argument", expr);
> > !       else
> > !         error ("invalid use of '%E' as a non-type 
> > template-argument", expr);
> > !
> >         return NULL_TREE;
> >       }
> >
> > --- 3085,3091 ----
> >       }
> >     else
> >       {
> > !       error ("object `%E' cannot be used as template argument", 
> > expr);
> >         return NULL_TREE;
> >       }
>
> People might get confused about the use of "object" here.  What do you
> think of something like
>
>       error ("`%E' cannot be used as a non-type template argument", 
> expr);

Again the patch is reversed so that error message is now gone.

Thanks,
Andrew Pinski