[patch] Fix PR 23439/23440, two alternatives

[Richard Henderson]

On Thu, Oct 06, 2005 at 01:31:13PM +0200, Volker Reichelt wrote:
> A couple of weeks ago I posted two alternatives to fix an
> ice-on-invalid-code of both the C and C++ frontend for code like this
> (see http://gcc.gnu.org/ml/gcc-patches/2005-09/msg00680.html ):
> 
>   void foo() { if ()
> 
> Both frontends call annotate_with_file_line with a NULL instead of a
> proper filename. The NULL is passed to strcmp which segfaults.

I think this misses the point that there's absolutely no reason we
should be using unknown_location.  If we've seen one token, then we
have a locus.

Why do you think unknown_location is relevant?


r~