Re: [diagnostics-branch] use precise locations in C front-end

[Chris Lattner <clattner at apple dot com>]

On Dec 12, 2008, at 11:05 AM, Aldy Hernandez wrote:
> > > Virtually all the C front-end is done, with the big exception of the
> > > build_fold*/fold* infrastructure.
> >
> > Nice, this sounds like good progress.  Out of curiosity, what do you
> > plan to do with fold?
>
> I haven't really thought about it, but something along the lines of  
> the
> routines taking a location and sticking that in the expression before
> returning the new expression.  But I'm sure it's more complicated than
> that (as is usually the case with GCC ;-)).

Ah ok, so you'll end up with tree nodes that have locations from the  
source code but that don't correspond to the source?

Here's a simple example: fold(x-x) -> 0.  What is the source location  
of the INTEGER_CST?

-Chris