FYI
-------- Forwarded Message --------
Subject: RE: complex.h functions and errno
Date: Thu, 9 Nov 2017 11:38:54 -0700
From: Barry Hedquist <b...@peren.com>
Reply-To: b...@peren.com
Organization: Perennial, Inc
To: 'Jim Thomas' <jaswtho...@sbcglobal.net>, 'David Keaton' <d...@dmk.com>
CC: 'Hal Finkel' <hfin...@anl.gov>
Hal,
From Jim Thomas. I’ll leave it to you to forward this to Austin Group.
Barry
*From:* Jim Thomas [mailto:jaswtho...@sbcglobal.net]
*Sent:* Thursday, November 09, 2017 11:17 AM
*To:* David Keaton <d...@dmk.com>
*Cc:* Barry Hedquist <b...@peren.com>
*Subject:* Re: complex.h functions and errno
On Nov 9, 2017, at 7:09 AM, David Keaton <d...@dmk.com
<mailto:d...@dmk.com>> wrote:
Barry,
I'm looping in Jim Thomas.
Hal is correct, and I think the response to him misses the
point that he raised.
Agreed.
What Hal was talking about was the fact that cabs(a + I*b) ==
sqrt(a*a + b*b), and the squaring done by cabs() is what causes the
overflow.
The C expressions cabs(a + I*b) and sqrt(a*a + b*b) aren’t equivalent
because the computation of the argument to sqrt overflows in many cases
where the corresponding cabs does not (e.g., a = DBL_MAX, b = 0, a+a +
b*b = inf, sqrt = inf, cabs = DBL_MAX). However, cabs can overflow in
extreme cases, like the one mentioned below: cabs(1.7e308+I*1.7e308).
Most of those who are interested in optimization would really
like for errno to go away, and I don't see any problem with POSIX
specifying that it isn't set for cases where the C standard leaves
it unspecified.
I think this gets to Hal’s point. The only problem I see (with the
C-POSIX difference) is for implementation of cabs(a + bi) as hypot(a, b)
if hypot sets errno. G.6#6 says
Each of the functions cabs and carg is specified by a formula in
terms of a real function (whose special cases are covered in annex F):
cabs(x + iy) = hypot(x, y)
carg(x + iy) = atan2(y, x)
Maybe one could argue that this specification does not apply to errno.
Jim
David
On 11/09/2017 06:52 AM, Barry Hedquist wrote:
Do we, WG14, have an answer for this? What does C11/C17 say
about this?
Thanks,
Barry
-----Original Message-----
From: Don Cragun [mailto:dcra...@sonic.net]
Sent: Wednesday, November 08, 2017 8:15 PM
To: Hal Finkel <hfin...@anl.gov <mailto:hfin...@anl.gov>>
Cc: austin-group-l <austin-group-l@opengroup.org
<mailto:austin-group-l@opengroup.org>>
Subject: Re: complex.h functions and errno
Hi Hal,
I agree with what others have already said in this thread, but I
think there
is a more basic issue at play here. In the example you provided:
cabs(1.7e308+I*1.7e308);
the multiplication is not being performed by the function. It
is performed
by the compiler evaluating arguments to be passed to the
functipn. So the
floating point overflow will occur before cabs() has a chance to
perform any
calculations.
The C99 standard (section 6.5 "Expressions") says:
"If an exceptional condition occurs during the evaluation of an
expression
(that is, if the result is not mathematically defined or not in the
range
of representable values for its type), the behavior is undefined."
which seems to be in conflict with the math error handling
requirements. (I
haven't tried to determine if this has been fixed in C11 or
later drafts.)
- Don
On Nov 7, 2017, at 3:22 PM, Hal Finkel <hfin...@anl.gov
<mailto:hfin...@anl.gov>> wrote:
Hi, everyone,
My reading of the combined C and POSIX specifications leads
me to believe
that functions in complex.h don't set errno on POSIX systems.
This came up
recently during a code review for the Clang compiler [1]. It has
been
pointed out that, under glibc, cabs() can set errno (e.g.,
cabs(1.7e308+I*1.7e308)) although POSIX says "No errors are
defined" for
cabs.
To elaborate, in the C specification, 7.12 specifies the
requirements for
functions in math.h. For those functions, 7.12.1 (Treatment of error
conditions) says that overflows do set ERANGE, and that it's
implementation
defined if the same is true for underflows. That's true for
functions in
math.h in general. 7.3 specifies the requirements for functions in
complex.h. Here, 7.3.2 says, "An implementation may set errno
but is not
required to." That, as I understand it, applies to all functions in
complex.h (unless otherwise noted, I presume). However, because
setting
errno is not required by C for functions in complex.h, when
POSIX says "No
errors are defined." that constrains the choice (POSIX is
constraining the
implementation choice allowed by C as it does in many other
cases). As a
result, under POSIX, the functions in complex.h don't set errno.
Is my reading correct? If it is correct, is it intentional?
I'd prefer
that these functions don't set errno, because that makes compiler
optimization easier, but my overriding concern is that the compiler
interprets the requirements correctly.
Thanks in advance,
Hal
[1] https://reviews.llvm.org/D39611
--
Hal Finkel
Lead, Compiler Technology and Programming Languages Leadership
Computing Facility Argonne National Laboratory
--
Hal Finkel
Lead, Compiler Technology and Programming Languages
Leadership Computing Facility
Argonne National Laboratory
