C++ PATCH to abate shift warnings (PR c++/68979)

Seems that people find compile-time error on the following testcase overly
pedantic.  I.e. that "enum A { X = -1 << 1 };" should compile, at least with
-fpermissive.  So I've changed the error_at into permerror and the return value
of cxx_eval_check_shift_p now depends on flag_permissive.  Luckily, I didn't
have to modify any of the existing tests.

Bootstrapped/regtested on x86_64-linux, ok for trunk?

2016-01-12  Marek Polacek  <polacek@redhat.com>

	PR c++/68979
	* constexpr.c (cxx_eval_check_shift_p): Use permerror rather than
	error_at and return negated flag_permissive.

	* g++.dg/warn/permissive-1.C: New test.


	Marek

On Tue, Jan 12, 2016 at 01:52:01PM +0100, Marek Polacek wrote:
> --- gcc/testsuite/g++.dg/warn/permissive-1.C
> +++ gcc/testsuite/g++.dg/warn/permissive-1.C
> @@ -0,0 +1,8 @@
> +// PR c++/68979
> +// { dg-do compile }
> +// { dg-options "-fpermissive -Wno-shift-overflow -Wno-shift-count-overflow -Wno-shift-count-negative" }
> +
> +enum A { AA = -1 << 4 }; // { dg-warning "operand of shift expression" "" { target c++11 } }
> +enum B { BB = 1 << -4 }; // { dg-warning "operand of shift expression" }
> +enum C { CC = 1 << 100 }; // { dg-warning "operand of shift expression" }
> +enum D { DD = 31 << 30 }; // { dg-warning "shift expression" "" { target c++11 } }

Shouldn't this test be limited to
// { dg-do compile { target int32 } }
or better yet replace the 100 and 30 above with
say __SIZEOF_INT__ * 4 * __CHAR_BIT__ - 4 and __SIZEOF_INT__ * __CHAR_BIT__ - 2
?
I'd guess that on say int16 targets, or int64 targets (if we have any at
some point) or int128 targets this wouldn't do what you are expecting.
{ target int32 } is not exactly right, because it still assumes __CHAR_BIT__ == 8
and for other char sizes it could fail.

	Jakub

Changing the diagnostic is OK, but cxx_eval_check_shift_p should return 
true regardless of flag_permissive, so that SFINAE results follow the 
standard.

Jason

On Tue, Jan 12, 2016 at 08:27:47AM -0500, Jason Merrill wrote:
> Changing the diagnostic is OK, but cxx_eval_check_shift_p should return true
> regardless of flag_permissive, so that SFINAE results follow the standard.

There's a complication, because if I keep returning true, we'll give a
compile-time error like this:

permissive-1.C:5:18: warning: left operand of shift expression ‘(-1 << 4)’ is
negative [-fpermissive]
 enum A { AA = -1 << 4 }; // { dg-warning "operand of shift expression" "" {
target c++11 } }

permissive-1.C:5:21: error: enumerator value for ‘AA’ is not an integer
constant
 enum A { AA = -1 << 4 }; // { dg-warning "operand of shift expression" "" {
target c++11 } }

So I suppose that wouldn't really help.  :(

	Marek

On 01/12/2016 09:05 AM, Marek Polacek wrote:
> On Tue, Jan 12, 2016 at 08:27:47AM -0500, Jason Merrill wrote:
>> Changing the diagnostic is OK, but cxx_eval_check_shift_p should return true
>> regardless of flag_permissive, so that SFINAE results follow the standard.
>
> There's a complication, because if I keep returning true, we'll give a
> compile-time error like this:
>
> permissive-1.C:5:18: warning: left operand of shift expression ‘(-1 << 4)’ is
> negative [-fpermissive]
>   enum A { AA = -1 << 4 }; // { dg-warning "operand of shift expression" "" {
> target c++11 } }
>
> permissive-1.C:5:21: error: enumerator value for ‘AA’ is not an integer
> constant
>   enum A { AA = -1 << 4 }; // { dg-warning "operand of shift expression" "" {
> target c++11 } }
>
> So I suppose that wouldn't really help.  :(

In that case, we need to return (!flag_permissive || ctx->quiet).

Jason