PR preprocessor/33305: Warn for empty macro arguments

[Joseph S. Myers]

On Fri, 13 Jun 2008, Andrew Haley wrote:

> Joseph S. Myers wrote:
> > Also, such a diagnostic should be restricted to -pedantic if your patch 
> > doesn't so restrict it, and needs testcases.
> 
> I can do that.  I don't agree that it should be restricted to -pedantic.
> Why do you think it should be?

It's a warning for the use of a GNU extension (a C99 feature supported as 
an extension in C90 mode), and such warnings are normally conditioned on 
-pedantic; we accept GNU extensions without warning by default.

-- 
Joseph S. Myers
joseph@codesourcery.com