Re: [fixinc] Fixing the last recalcitrant test case

[Bruce Korb <bkorb at veritas dot com>]

Nathanael Nerode wrote:
> 
> This issue consists entirely of line endings.  The 'undefine_null' hack
> tests for \r\n terminated lines, *and preserves the \r\n ending*.  This is
> tested. This means that the actual output after the hack always has \r\n
> line endings.  But the base copy doesn't.
> 
> I could amend the base copy to have the appropriate \r\n line endings (as
> in the pseudo-patch below).  But I'm not at all sure that's the right thing
> to do; I doubt that we really want that in CVS or tarballs or anything.
> Since there are *lots* of other hacks which insert lines with \n endings,
> I don't see why we're preserving the \r\n line endings in this case.  So
> perhaps I could simply change the fix to drop the \r's it finds.
> 
> Any opinions on what is best here?

No.  :-)  Looking at this cruft, it looks like I was preserving
some behavior from way back whenever.  It is a nuisance.  If the
``-qwb'' option makes this issue go away, then forget it.  Otherwise,
we should either stop messing with \r or we should fix the base test.

> Index: tests/base/testing.h
> ===================================================================
> RCS file: /cvs/gcc/gcc/gcc/fixinc/tests/base/testing.h,v
> retrieving revision 1.12
> diff -u -r1.12 testing.h
> --- tests/base/testing.h        22 Apr 2003 23:41:28 -0000      1.12
> +++ tests/base/testing.h        26 Aug 2003 23:51:56 -0000
> @@ -113,9 +113,9 @@
> 
> 
>  #if defined( UNDEFINE_NULL_CHECK )
> -#ifndef NULL
> -#define NULL 0UL
> -#endif
> +#ifndef NULL
> +#define NULL 0UL
> +#endif
>  #ifndef NULL
>  #define NULL   ((void*)0)
>  #endif