public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug fortran/19904] New: Division by zero leads to error
@ 2005-02-11 17:33 schnetter at aei dot mpg dot de
2005-02-11 19:16 ` [Bug fortran/19904] " pinskia at gcc dot gnu dot org
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: schnetter at aei dot mpg dot de @ 2005-02-11 17:33 UTC (permalink / raw)
To: gcc-bugs
I want to define a parameter that contains a nan. When I use a division by
zero in a parameter statement, gfortran aborts with an error instead of
setting the parameter to nan. (The Intel Fortran compiler accepts this
parameter statement without warning.)
subroutine x
implicit none
double precision, parameter :: zero=0, nan=0/zero
end subroutine x
$ ~/gcc/bin/gfortran -c zero.f90
In file zero.f90:3
double precision, parameter :: zero=0, nan=0/zero
1
Error: Division by zero at (1)
$ ~/gcc/bin/gfortran --version
GNU Fortran 95 (GCC 4.0.0 20050126 (experimental))
--
Summary: Division by zero leads to error
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: fortran
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: schnetter at aei dot mpg dot de
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: i686-pc-linux-gnu
GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread* [Bug fortran/19904] Division by zero leads to error
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
@ 2005-02-11 19:16 ` pinskia at gcc dot gnu dot org
2005-02-11 21:34 ` sgk at troutmask dot apl dot washington dot edu
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2005-02-11 19:16 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-11 13:50 -------
Confirmed.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed| |1
Keywords| |rejects-valid
Last reconfirmed|0000-00-00 00:00:00 |2005-02-11 13:50:33
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread* [Bug fortran/19904] Division by zero leads to error
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
2005-02-11 19:16 ` [Bug fortran/19904] " pinskia at gcc dot gnu dot org
@ 2005-02-11 21:34 ` sgk at troutmask dot apl dot washington dot edu
2005-02-11 21:50 ` schnetter at aei dot mpg dot de
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: sgk at troutmask dot apl dot washington dot edu @ 2005-02-11 21:34 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From sgk at troutmask dot apl dot washington dot edu 2005-02-11 17:10 -------
Gfortran is a Fortran 95 compiler, and AFAIK gfortran's behavior conforms
to the Fortran 95 standard, because it has no allowance for floating point
exception. Someday, gfortran may/will support Technical Report ISO/IEC
15580: 1998(E), which will then permit the use of the ieee_arithmetic,
ieee_features, and ieee_exceptions intrinsic modules.
If you want to set a NaN in gfortran you can do,
program test
real nan
integer duh
equivalence (nan,duh)
duh = z'FFC00000'
print*, nan
end program test
The question then becomes "what are going to do with NaN?"
I suspect the gfortran's behavior of the numeric rational
operation and numeric binary operation may give undefined
behavior, which is again permitted by the Fortran 95 standard.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread* [Bug fortran/19904] Division by zero leads to error
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
2005-02-11 19:16 ` [Bug fortran/19904] " pinskia at gcc dot gnu dot org
2005-02-11 21:34 ` sgk at troutmask dot apl dot washington dot edu
@ 2005-02-11 21:50 ` schnetter at aei dot mpg dot de
2005-02-11 22:48 ` sgk at troutmask dot apl dot washington dot edu
2005-02-11 23:42 ` pinskia at gcc dot gnu dot org
4 siblings, 0 replies; 6+ messages in thread
From: schnetter at aei dot mpg dot de @ 2005-02-11 21:50 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From schnetter at aei dot mpg dot de 2005-02-11 17:17 -------
Actually, gfortran handles nans correctly (according to the IEEE standard) in
all other cases. I can e.g. generate a nan through the expression a/b, if I
only hide from the compiler the fact that b is zero, so that it generates code
that performs the division at run time, and does not try to evaluate it at
compile time.
Also, while your argument is correct that the Fortran 95 standard does not
require nans to be handled correctly, your code is also undefined according to
the Fortran 95 standard, because you use the variable "nan" without defining
it: if you assign to an equivalenced storage an integer value, then you are
only allowed to read out the integer value, and vice versa for the real value.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug fortran/19904] Division by zero leads to error
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
` (2 preceding siblings ...)
2005-02-11 21:50 ` schnetter at aei dot mpg dot de
@ 2005-02-11 22:48 ` sgk at troutmask dot apl dot washington dot edu
2005-02-11 23:42 ` pinskia at gcc dot gnu dot org
4 siblings, 0 replies; 6+ messages in thread
From: sgk at troutmask dot apl dot washington dot edu @ 2005-02-11 22:48 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From sgk at troutmask dot apl dot washington dot edu 2005-02-11 17:44 -------
You are probably right that my example does not conform to the standard,
because I don't use equilance() and I whipped up the example in a minute
without inspection of its conformance.
Anyway, if you want to make gfortran accept your code, constant folding
is done in arith.c. I suggest that you add a -fallow_inf_nan option
to gfortran and make the appropriate modifications to arith.c
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread* [Bug fortran/19904] Division by zero leads to error
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
` (3 preceding siblings ...)
2005-02-11 22:48 ` sgk at troutmask dot apl dot washington dot edu
@ 2005-02-11 23:42 ` pinskia at gcc dot gnu dot org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2005-02-11 23:42 UTC (permalink / raw)
To: gcc-bugs
--
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |enhancement
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19904
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2005-02-11 20:32 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-02-11 17:33 [Bug fortran/19904] New: Division by zero leads to error schnetter at aei dot mpg dot de
2005-02-11 19:16 ` [Bug fortran/19904] " pinskia at gcc dot gnu dot org
2005-02-11 21:34 ` sgk at troutmask dot apl dot washington dot edu
2005-02-11 21:50 ` schnetter at aei dot mpg dot de
2005-02-11 22:48 ` sgk at troutmask dot apl dot washington dot edu
2005-02-11 23:42 ` pinskia at gcc dot gnu dot org
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).