Re: Integer constant parsing overhaul

[Zack Weinberg <zackw at panix dot com>]

On Thu, Aug 23, 2001 at 06:39:16PM +0100, Neil Booth wrote:
> Hi Zack,
> 
> This looks good.  I have a couple of nits that you may or may not
> agree with.
[...]
> > +    handling in GCC proper is too ugly to speak of.  */
> 
> Maybe this should be moved to above cpp_classify_number(), which has
> no comment immediately above it.

Good idea...

> How about scrapping this last if statement and its block, and putting
> something like the following in the switch?
> 
>        case 'l': case 'L':
> 	 if (l++ && s[n - 1] != s[n - 2])
> 	   return 0;
> 	 break;
> 
> Then you can lose save_n too.  This might written a little more
> cleanly by decrementing n as part of the loop condition.

You're right that the way it's written is confusing, but I am not
sure whether your way is an improvement - particularly in view of
your second message about this.  I'll see what I can do.

> > + unsigned HOST_WIDEST_INT
> > + cpp_interpret_integer (pfile, tok, flags)
> 
> No comment for this function.

Argh.

> > !   for (; itk < itk_none; itk += 2 /* skip signed types */)
> > !     if (int_fits_type_p (value, integer_types[itk]))
> > !       return itk;
> 
> Skip unsigned types?

Yep.  Cut-and-paste error.

> Some of the comments in the code don't really
> follow the GNU coding standards.

These inside-the-loop-expression comments would be less helpful
if written as complete sentences on a separate line, IMO.  Where I
put them, it's quite clear what they are explaining.

zw