On Sun, Mar 30, 2014 at 11:05 AM, Nico Rieck <[email protected]> wrote:
> On 30.03.2014 19:35, Reid Kleckner wrote:
> > Should we be more specific here? Something like:
> > redeclaration %0 adds attribute %1 which was missing the previous
> > declaration
>
> Sounds like a good idea. I would phrase it: "redeclaration of %q0 may
> not add %q1 attribute" (or rather "cannot"?) which would look like:
>
> t.cpp:222:28: error: redeclaration of 'X::f' may not add 'dllexport'
> attribute
> __declspec(dllexport) void X::f() {}
> ^
> t.cpp:219:6: note: previous declaration is here
> void f();
>
I like your wording better. I'd go with "cannot". DiagnosticSemaKinds.td
has 28 "may not" instances and ~370 "cannot" instances.
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
