Re: wide-int

[Kenneth Zadeck <zadeck at naturalbridge dot com>]

On 11/22/2013 03:03 PM, Richard Biener wrote:
> On Thu, Nov 21, 2013 at 11:08 PM, Mike Stump <mikestump@comcast.net> wrote:
> > This patch adds support for ints wider than double_int.
> >
> > Ok?
> Please split the patch into pieces.  I suggest to separate changes to
> the various frontends
> (CC maintainers), the new wide-int files, the example rs6000
> conversion and the rest.
>
> Otherwise I'm sure you'll get no attention.
we are happy to break the patch any way that makes since.

> Otherwise I'm sure you need to do some teaching about the different
> wide-int kinds
> and their behavior.
i will start writing some documentation.
> Is the patch an old revision?  Because I see the int_const_binop_1
> code that I asked
> Kenny to revert to trunk behavior (return TREE_OVERFLOW flagged
> constants, not NULL_TREE).
this is a special case.    what we did with the stor-layout patch was 
that we committed the fix the the branch simultaneously with the fix 
going on the trunk.

i have the fix for the wide-int branch, i was waiting for you to approve 
the fix for trunk.
(as a side point i think that their is a third test case that i am going 
to have fix for it.   it is not c or c++   i have a build with every 
language going on now to check this. )

> +#ifndef WIDE_INT_PRINT_H
> +#define WIDE_INT_PRINT_H
> +
> +#include <stdio.h>
> +#include "wide-int.h"
> +
>
> system headers must be only included from [t]system.h
>
> For the merge you want to have trunk frozen and you want to be
> available during the following
> days (I do expect fallout).
unless you want to go the route of approving the wide-int* first, the 
rtl second ....   But yes, i think that we would request a sunday to do 
the commit and be around for the next few days.

kenny

> Thanks,
> Richard.