Re: your recent patch to cpplib.c

[Zack Weinberg <zack at wolery dot cumb dot org>]

On Tue, Feb 08, 2000 at 05:12:59PM -0800, Mark Mitchell wrote:
> 
>   I didn't see a gcc-patches message for this:
> 
> I apologize for that.  Our mail server was very badly behaved
> yesterday -- it dropped outgoing mail without telling me!  I'm still
> trying to figure out who got what.

Ah, that would explain it.  Thanks for clarifying.

> I posted a test-case with the bug.  It was, I belive:
> 
>   #define f(x) x
>   void g() {
>     f 
>       (3);
>   }
> 
> The CPP output compressed `g' to one line.

Right.  Contrariwise, consider

   #define f(x) x
   void g() {
     (f
       )(3);
   }

That will come out with an extra line with your patch, but not the way
it used to be.

>   The right fix is not to use the mark at all in this context.  It's
>   supposed to be for lookahead under extremely limited circumstances and
>   this isn't one of them.  (And it will probably vanish entirely in the
>   parser rewrite.)
> 
>   On the other hand, congratulations; you have probably found the line
>   numbering bug I've been looking for for months.
> 
> Well, that's something.  I apologize for the bogus fix.  What should
> we do at this point?
> 
> I've not got the time to do the complete fix, I don't think.  But, I
> can certainly back out my change, or we can leave it in.  Let me know
> what you would like.

Let's leave it in.  The bug is now in the rarer case, which is better
than nothing.  I'm working on a patch that will fix this as a side
effect, but it touches a bunch of other code and needs more testing
before I can publish it.

> And thanks for explaining this.

No problem.

zw