PATCH to test suite for better tokenization

[Joseph S. Myers]

On Sat, 12 Mar 2005, Per Bothner wrote:

> Zack Weinberg wrote:
> > OK, provided that this passes both with and without
> > --enable-mapped-location.
> 
> It does that.
> 
> I did see a weird regression in gcc.dg/asm-wide-1.c for
> the --enable-mapped-location case.  It's weird, because the errors
> look identical to the passing case.  I'll take a closer look some
> other time.

I was never really convinced a lot of the errors there were on the ideal 
tokens (in particular the wide strings should ideally be on their own 
lines in the testcase and the "wide string literal in 'asm'" errors on the 
lines of the wide strings).  The other errors aren't of particular 
importance but in general errors from a string in an asm should be on that 
string token and I suppose others should be on some sensible location for 
the asm as a whole.  Feel free to refine location of asm diagnostics and 
move tokens in that testcase onto their own lines if it's necessary to get 
consistent results with / without --enable-mapped-location.

-- 
Joseph S. Myers               http://www.srcf.ucam.org/~jsm28/gcc/
    jsm@polyomino.org.uk (personal mail)
    joseph@codesourcery.com (CodeSourcery mail)
    jsm28@gcc.gnu.org (Bugzilla assignments and CCs)