[Bug libc/6981] __STDC_IEC_559__ should not be defined unconditionally

["vincent-srcware at vinc17 dot net" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=6981

--- Comment #9 from Vincent Lefèvre <vincent-srcware at vinc17 dot net> ---
(In reply to joseph@codesourcery.com from comment #8)
> There would be no reason for F.6 and F.10.11 (in C11) if the intent wasn't 
> to support evaluation in wider types.

The fact that both F.6 and F.10.11 are new in C11 shows that this is pointless
in C99. And I disagree about F.6: you can have

double foo (void)
{
  return LDBL_EPSILON;
}

int main (void)
{
  double x = 17.0L * foo ();
}

I don't think that C99 mandates that LDBL_EPSILON be converted to double if the
eval method is implementation-defined (the arithmetic operations would be
correctly rounded as required by IEEE 754, possibly with an extended exponent
range, but the implementation could decide for C specific operations such as
function calls). F.6 in C11 makes the behavior explicit.

Concerning F.10.11, there's a contradiction with some codes given earlier in
Annex F, which assume FLT_EVAL_METHOD to be 0 or a value with
implementation-defined behavior.

Now, it is not even clear that the following would be forbidden:
FLT_EVAL_METHOD is 1 or 2, but rounding for the arithmetic operations is done
in the semantic type. For instance, I suspect that FLT_EVAL_METHOD should be 2
with the x87 FPU, even when configured to round to double precision, because
the intermediate values still have the extended exponent range (unless the
compiler chooses to store the intermediate values to memory). In the C
standard, FLT_EVAL_METHOD = 2 doesn't imply that the full extended precision is
used, just that the intermediate values are representable in long double.

> DTS 18661-1 clarifies things further [...]

Well, it's more than clarification, it is based on IEEE 754-2008, whose
requirements are different from the old IEEE 754-1985 standard: the old
requirement I mentioned became obsoleted (replaced) by the use of formatOf
operations. But C99 and C11's Annex F is based on the old IEEE 754-1985
standard.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
>From glibc-bugs-return-21356-listarch-glibc-bugs=sources.redhat.com@sourceware.org Sat Feb 08 00:11:03 2014
Return-Path: <glibc-bugs-return-21356-listarch-glibc-bugs=sources.redhat.com@sourceware.org>
Delivered-To: listarch-glibc-bugs@sources.redhat.com
Received: (qmail 15604 invoked by alias); 8 Feb 2014 00:11:03 -0000
Mailing-List: contact glibc-bugs-help@sourceware.org; run by ezmlm
Precedence: bulk
List-Id: <glibc-bugs.sourceware.org>
List-Subscribe: <mailto:glibc-bugs-subscribe@sourceware.org>
List-Post: <mailto:glibc-bugs@sourceware.org>
List-Help: <mailto:glibc-bugs-help@sourceware.org>, <http://sourceware.org/lists.html#faqs>
Sender: glibc-bugs-owner@sourceware.org
Delivered-To: mailing list glibc-bugs@sourceware.org
Received: (qmail 15580 invoked by uid 55); 8 Feb 2014 00:10:59 -0000
From: "joseph at codesourcery dot com" <sourceware-bugzilla@sourceware.org>
To: glibc-bugs@sourceware.org
Subject: [Bug libc/6981] __STDC_IEC_559__ should not be defined unconditionally
Date: Sat, 08 Feb 2014 00:11:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: glibc
X-Bugzilla-Component: libc
X-Bugzilla-Version: unspecified
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: joseph at codesourcery dot com
X-Bugzilla-Status: RESOLVED
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at sourceware dot org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-6981-131-cvhmhwGwTD@http.sourceware.org/bugzilla/>
In-Reply-To: <bug-6981-131@http.sourceware.org/bugzilla/>
References: <bug-6981-131@http.sourceware.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://sourceware.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-02/txt/msg00333.txt.bz2
Content-length: 1816

http://sourceware.org/bugzilla/show_bug.cgi?idi81

--- Comment #10 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
On Fri, 7 Feb 2014, vincent-srcware at vinc17 dot net wrote:

> Concerning F.10.11, there's a contradiction with some codes given earlier in
> Annex F, which assume FLT_EVAL_METHOD to be 0 or a value with
> implementation-defined behavior.

I don't see what problems you see with any example code in Annex F (but in
any case, if there are contradictions - which you should raise as C11 DRs
if so - it's natural to try to resolve them in accordance with the clear
intent that FLT_EVAL_METHOD being 2 is consistent with Annex F.

> Now, it is not even clear that the following would be forbidden:
> FLT_EVAL_METHOD is 1 or 2, but rounding for the arithmetic operations is done
> in the semantic type. For instance, I suspect that FLT_EVAL_METHOD should be 2
> with the x87 FPU, even when configured to round to double precision, because
> the intermediate values still have the extended exponent range (unless the
> compiler chooses to store the intermediate values to memory). In the C
> standard, FLT_EVAL_METHOD = 2 doesn't imply that the full extended precision is
> used, just that the intermediate values are representable in long double.

"evaluate all operations and constants to the range and precision of the
long double type" seems clear enough to me.

If the x87 FPU is rounding to double precision, that's a different ABI, as
used on 32-bit FreeBSD, where long double has the extended-precision
in-memory encoding but only 53 mantissa bits are used; that ABI isn't
supported by glibc (calling libm functions with the FPU rounding to double
precision means undefined behavior).

--
You are receiving this mail because:
You are on the CC list for the bug.